co-occurrence_of_enzyme_domains_(main).ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Introduction \n",
    "You will want to run the first two cells no matter what. After that, the notebook is split up into sections dedicated to generating specific figures (and supporting figures) of the manuscript."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "###settings used thoughout the entire notebook:\n",
    "import pandas\n",
    "\n",
    "pfam_names_and_descriptions_filename=\"pfamID_to_name_desc_longdesc.tsv\"\n",
    "annotation_categories_we_care_about=[\"y\",\"s\",\"e\",\"m\"]\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "pfam_id_name_description_df=pandas.read_table(pfam_names_and_descriptions_filename,index_col=0,dtype=\"str\",keep_default_na=False, na_values=[])\n",
    "\n",
    "#grabbing domains we care about from the descriptions file\n",
    "domains_we_care_about=pfam_id_name_description_df[pfam_id_name_description_df[\"enzyme_annotation\"].isin(annotation_categories_we_care_about)].index.to_list()\n",
    "\n",
    "\n",
    "# loading domains from a file\n",
    "# domains_we_care_about_filename=\"enzymatic_domains.txt\"\n",
    "# with open(domains_we_care_about_filename,'r') as file:\n",
    "#     domains_we_care_about=file.read().splitlines()\n",
    "\n",
    "\n",
    "# if you wanted to look at ALL pfam domains (works for analyzing PfamScan output but probably won't work for all-of-uniprot analysis)\n",
    "# domains_we_care_about=[\"PF{:05d}\".format(i) for i in range(20000)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Completed\n"
     ]
    }
   ],
   "source": [
    "###some setup we need for this notebook in general\n",
    "import itertools\n",
    "import collections\n",
    "import networkx\n",
    "import json\n",
    "import re\n",
    "\n",
    "###define some functions\n",
    "       \n",
    "def load_PfamScan_results_dataframe(input_filename,verbose=True):\n",
    "    if(verbose):\n",
    "        print(\"loading \"+input_filename)\n",
    "    pfamscan_results_df=pandas.read_csv(input_filename,\n",
    "                                          delim_whitespace=True,\n",
    "#                                           dtype=\"str\",\n",
    "                                          comment=\"#\",\n",
    "                                          header=None,\n",
    "                                          usecols=[0,5,7,12,14],\n",
    "                                          names=[\"protein_id\",\"pfam_accession\",\"type\",\"evalue\",\"clan\"])\n",
    "\n",
    "    #get rid of the version part of the pfam accession\n",
    "    pfamscan_results_df[\"pfam_accession\"]=pfamscan_results_df[\"pfam_accession\"].str.split(\".\").str[0]\n",
    "    \n",
    "    return pfamscan_results_df\n",
    "\n",
    "\n",
    "\n",
    "def keep_only_lowest_evalue_per_clan(dataframe,verbose=True):\n",
    "    \"\"\"\n",
    "    pfam_scan removes domains in the same clan that are overlapping, but sometimes it doesn't detect the\n",
    "    overlap quite right. Since we mostly care about the co-ocurrence of domains that are quite different,\n",
    "    we just throw out all domains in the same clan in the same protein, whether they overlap or not.\n",
    "    this function takes care of that.\n",
    "    \"\"\"\n",
    "    list_of_indexes_we_want_to_keep=[]\n",
    "    counter=0\n",
    "    grouped_df=dataframe.groupby(\"protein_id\")\n",
    "    if (verbose):\n",
    "        print(\"filtering for best hit per clan \")\n",
    "        total_groups=len(grouped_df)\n",
    "        count=0\n",
    "    for protein_id,protein_grouped_df, in grouped_df:\n",
    "        if (verbose):\n",
    "            if(count%1000==0 and count>0):\n",
    "                print(\"now processing protein #{} out of {}\".format(count,total_groups))\n",
    "            count+=1\n",
    "        for clan,clan_grouped_df in protein_grouped_df.groupby(\"clan\"):\n",
    "            if(clan==\"No_clan\"):\n",
    "                list_of_indexes_we_want_to_keep+=clan_grouped_df.index.to_list()\n",
    "            else:\n",
    "                list_of_indexes_we_want_to_keep.append(clan_grouped_df[\"evalue\"].idxmin())\n",
    "    return dataframe.loc[list_of_indexes_we_want_to_keep]\n",
    "\n",
    "\n",
    "\n",
    "def load_PfanScan_results_only_domains_we_care_about_dataframe(input_filename,verbose=True):\n",
    "    pfamscan_results_df=load_PfamScan_results_dataframe(input_filename,verbose)\n",
    "    if (verbose):\n",
    "        print(\"filtering for domains we care about\")\n",
    "    pfamscan_results_with_domains_we_care_about=pfamscan_results_df[pfamscan_results_df[\"pfam_accession\"].isin(domains_we_care_about)]\n",
    "    return keep_only_lowest_evalue_per_clan(pfamscan_results_with_domains_we_care_about,verbose)\n",
    "\n",
    "\n",
    "\n",
    "def load_PfamScan_domain_pairs_we_care_about(input_filename,verbose=True):\n",
    "    filtered_pfamscan_results_df=load_PfanScan_results_only_domains_we_care_about_dataframe(input_filename,verbose)\n",
    "    domain_to_protein_list_dict=collections.defaultdict(list)\n",
    "    domain_pair_to_protein_list_dict=collections.defaultdict(list)\n",
    "    grouped_df=filtered_pfamscan_results_df.groupby(\"protein_id\")\n",
    "    for protein_id, data_series in grouped_df:\n",
    "        domain_entries=frozenset(data_series[\"pfam_accession\"]) \n",
    "        for entry in domain_entries:\n",
    "            domain_to_protein_list_dict[entry].append(protein_id)\n",
    "        for combination in itertools.combinations(domain_entries,r=2):\n",
    "            domain_pair_to_protein_list_dict[frozenset(combination)].append(protein_id)\n",
    "    return domain_to_protein_list_dict,domain_pair_to_protein_list_dict\n",
    "\n",
    "print(\"Completed\")\n",
    "\n",
    "\n",
    "\n",
    "#to use this function, the dataframe needs a \"Cross-reference (Pfam)\" column and the protein IDs need to be the index\n",
    "def preprocess_domain_pairs_in_uniprot_dataframe(df,verbose=True):\n",
    "    domain_mapping_list_dict=collections.defaultdict(list)\n",
    "    domain_pair_frozenset_list_dict=collections.defaultdict(list)\n",
    "    counter=0\n",
    "    for protein_id,row in df.iterrows():\n",
    "        if(counter%10000==0 and counter>0 and verbose):\n",
    "            print(\"now processing row {} of {}\".format(counter,len(df)))\n",
    "        counter+=1\n",
    "        pfams_string=row[\"Cross-reference (Pfam)\"]\n",
    "        entries=frozenset(pfams_string.split(\";\")[:-1]) #the \"[:-1]\" is to drop the last element, which is always empty because the strings end in \";\"\n",
    "        entries=entries.intersection(domains_we_care_about)\n",
    "        for entry in entries:\n",
    "            domain_mapping_list_dict[entry].append(protein_id)\n",
    "        for combination in itertools.combinations(entries,r=2):\n",
    "            domain_pair_frozenset_list_dict[frozenset(combination)].append(protein_id)\n",
    "    return domain_mapping_list_dict,domain_pair_frozenset_list_dict\n",
    "\n",
    "def test_simplestring(string):\n",
    "    assert isinstance(string,str), \"\\\"{}\\\" needs to be a string\".format(string)\n",
    "    assert re.match(r'^[A-Za-z0-9_-]+$', string), \"\\\"{}\\\" may only contain alphanumeric characters, dash and underscore\".format(string)\n",
    "\n",
    "class DomainOccurrences:\n",
    "    def __init__(self, name_string, domain_to_protein_list_dict, domain_pair_frozenset_to_protein_list_dict):\n",
    "        assert isinstance(domain_to_protein_list_dict,dict), \"domain_to_protein_list_dict needs to be a dict\"\n",
    "        assert isinstance(domain_pair_frozenset_to_protein_list_dict,dict), \"domain_pair_frozenset_to_protein_list_dict needs to be a dict\"\n",
    "        test_simplestring(name_string)\n",
    "        #make sure that all domains that have pairs are also in the single-domain dict\n",
    "        for domain in domain_to_protein_list_dict:\n",
    "            test_simplestring(domain)\n",
    "        for domain_pair in domain_pair_frozenset_to_protein_list_dict:\n",
    "            for domain in domain_pair:\n",
    "                test_simplestring(domain)\n",
    "                assert domain in domain_to_protein_list_dict, \"domain {} not found in domain_to_protein_list_dict\".format(domain)\n",
    "        \n",
    "        self.domain_to_protein_list_dict=dict(domain_to_protein_list_dict)#!\n",
    "        self.domain_pair_frozenset_to_protein_list_dict=dict(domain_pair_frozenset_to_protein_list_dict)#!\n",
    "        self.name=name_string\n",
    "    \n",
    "    def get_domain_protein_list(self,entry):\n",
    "        if entry in self.domain_to_protein_list_dict:\n",
    "            return list(self.domain_to_protein_list_dict[entry])\n",
    "        else:\n",
    "            return []\n",
    "    \n",
    "    def get_domain_protein_tally(self,entry):\n",
    "        if entry in self.domain_to_protein_list_dict:\n",
    "            return len(self.domain_to_protein_list_dict[entry])\n",
    "        else:\n",
    "            return 0\n",
    "\n",
    "    def get_domain_pair_protein_list(self,entries_frozenset):\n",
    "        if entries_frozenset in self.domain_pair_frozenset_to_protein_list_dict:\n",
    "            return list(self.domain_pair_frozenset_to_protein_list_dict[entries_frozenset])\n",
    "        else:\n",
    "            return []\n",
    "        \n",
    "    def get_domain_pair_protein_tally(self,entries_frozenset):\n",
    "        if entries_frozenset in self.domain_pair_frozenset_to_protein_list_dict:\n",
    "            return len(self.domain_pair_frozenset_to_protein_list_dict[entries_frozenset])\n",
    "        else:\n",
    "            return 0\n",
    "\n",
    "    def generate_coocurrence_network(self,\n",
    "                                     output_filename,\n",
    "                                     *secondary_domaincooccurrences, #zero or more\n",
    "                                     generate_directed_graph=True,\n",
    "                                     draw_node_when=lambda occurrence: occurrence>=1,\n",
    "                                     draw_edge_when=lambda absolute_coocurrence,pct_coocurrence: absolute_coocurrence>=1, #pct_cooccurrence is always at 100 for generate_directed_graph=True\n",
    "                                     max_entries_in_list=100,\n",
    "                                     verbose=True):\n",
    "\n",
    "        truncate_list=lambda input_list: input_list if len(input_list)<=max_entries_in_list else [\"Too many entries, only showing \"+str(max_entries_in_list)+\" (pass a max_entries_in_list argument to change this)\"]+input_list[0:max_entries_in_list-1]\n",
    "\n",
    "        def generate_edge_properties(accessions_list,entries_frozenset):\n",
    "            cooccurring_domains_in_mibig_protein_list=mibig_domainoccurrences.get_domain_pair_protein_list(entries_frozenset)\n",
    "            cooccurring_domains_in_mibig_protein_list_with_accessions=[k+\":\"+v for k,v in uniprot_to_mibig_mapping_df.loc[cooccurring_domains_in_mibig_protein_list,\"mibig_accession\"].iteritems()]\n",
    "            cooccurring_domains_in_uniprot_curated_catalytic_list=uniprot_curated_catalytic_domainoccurrences.get_domain_pair_protein_list(entries_frozenset)\n",
    "            cooccurring_domains_in_uniprot_curated_catalytic_list_with_rhea_ids=[k+\":\"+v for k,v in uniprot_curated_catalytic_df.loc[cooccurring_domains_in_uniprot_curated_catalytic_list,\"Rhea Ids\"].iteritems()]\n",
    "            cooccurring_domains_in_brenda_list=brenda_domainoccurrences.get_domain_pair_protein_list(entries_frozenset)\n",
    "            cooccurring_domains_in_brenda_list_with_accessions=[k+\":\"+v for k,v in uniprot_to_brenda_mapping_df.loc[cooccurring_domains_in_brenda_list,\"brenda_ec_number\"].iteritems()]\n",
    "            tally=len(accessions_list)\n",
    "\n",
    "            in_databases_abbreviation_list=[]\n",
    "            in_databases_superscript_abbreviation_list=[]\n",
    "            if len(cooccurring_domains_in_uniprot_curated_catalytic_list)>0:\n",
    "                in_databases_abbreviation_list.append(\"U\")\n",
    "                in_databases_superscript_abbreviation_list.append(\"ᵁ\")\n",
    "            if len(cooccurring_domains_in_mibig_protein_list)>0:\n",
    "                in_databases_abbreviation_list.append(\"M\")\n",
    "                in_databases_superscript_abbreviation_list.append(\"ᴹ\")\n",
    "            if len(cooccurring_domains_in_brenda_list)>0:\n",
    "                in_databases_abbreviation_list.append(\"B\")\n",
    "                in_databases_superscript_abbreviation_list.append(\"ᴮ\")\n",
    "\n",
    "            edge_properties_dict={\n",
    "                \"cooccurrences_list\":truncate_list(accessions_list),\n",
    "                \"cooccurrences_count\":tally,\n",
    "\n",
    "                \"cooccurring_domains_in_uniprot_curated_catalytic_list\":truncate_list(cooccurring_domains_in_uniprot_curated_catalytic_list),\n",
    "                \"cooccurring_domains_in_uniprot_curated_catalytic_list_with_rhea_ids\":truncate_list(cooccurring_domains_in_uniprot_curated_catalytic_list_with_rhea_ids),\n",
    "                \"cooccurring_domains_in_uniprot_curated_catalytic_count\":len(cooccurring_domains_in_uniprot_curated_catalytic_list),\n",
    "                \"cooccurring_domains_in_uniprot_curated_catalytic_boolean\":len(cooccurring_domains_in_uniprot_curated_catalytic_list)>0,\n",
    "\n",
    "                \"cooccurring_domains_in_mibig_protein_list\":truncate_list(cooccurring_domains_in_mibig_protein_list),\n",
    "                \"cooccurring_domains_in_mibig_protein_list_with_mibig_accessions\":truncate_list(cooccurring_domains_in_mibig_protein_list_with_accessions),\n",
    "                \"cooccurring_domains_in_mibig_protein_count\":len(cooccurring_domains_in_mibig_protein_list),\n",
    "                \"cooccurring_domains_in_mibig_protein_boolean\":len(cooccurring_domains_in_mibig_protein_list)>0,\n",
    "\n",
    "                \"cooccurring_domains_in_brenda_list\":truncate_list(cooccurring_domains_in_brenda_list),\n",
    "                \"cooccurring_domains_in_brenda_list_with_ec_numbers\":truncate_list(cooccurring_domains_in_brenda_list_with_accessions),\n",
    "                \"cooccurring_domains_in_brenda_count\":len(cooccurring_domains_in_brenda_list),\n",
    "                \"cooccurring_domains_in_brenda_boolean\":len(cooccurring_domains_in_brenda_list)>0,\n",
    "\n",
    "                \"cooccurring_domains_in_databases_abbreviations\":\"\".join(in_databases_abbreviation_list),\n",
    "                \"cooccurring_domains_in_databases_abbreviation_list\":in_databases_abbreviation_list,\n",
    "\n",
    "                \"human_readable_string_a\":str(tally),\n",
    "                \"human_readable_string_b\":str(tally)+(\" proteins\" if tally>1 else \" protein\"),\n",
    "                \"human_readable_string_c\":str(tally)+\" \"+\"\".join([\"(\"+x+\")\" for x in in_databases_abbreviation_list]),\n",
    "                \"human_readable_string_d\":str(tally)+\" (\"+\"\".join(in_databases_abbreviation_list)+\")\",\n",
    "                \"human_readable_string_e\":str(tally)+(\" proteins\" if tally>1 else \" protein\")+\"\\n\"+\"\".join([\"(\"+x+\")\" for x in in_databases_abbreviation_list]),\n",
    "                \"human_readable_string_f\":str(tally)+\"\".join(in_databases_superscript_abbreviation_list),\n",
    "                \"human_readable_string_g\":str(tally)+(\"⁽\"+\"\".join(in_databases_superscript_abbreviation_list)+\"⁾\" if len(in_databases_superscript_abbreviation_list)>0 else \"\"),\n",
    "                \"human_readable_string_h\":str(tally)+\" \"+\"\".join([\"⁽\"+x+\"⁾\" for x in in_databases_superscript_abbreviation_list]),\n",
    "                \"human_readable_string_i\":str(tally),\n",
    "                \"human_readable_string_j\":str(tally)+(\" proteins\" if tally>1 else \" protein\"),\n",
    "            }\n",
    "            \n",
    "            for domainoccurrence in secondary_domaincooccurrences:\n",
    "                domainoccurrence_list=domainoccurrence.get_domain_pair_protein_list(entries_frozenset) #!\n",
    "                domainoccurrence_len=len(domainoccurrence_list)\n",
    "                edge_properties_dict[\"cooccurring_domains_in_\"+domainoccurrence.name+\"_list\"]=truncate_list(domainoccurrence_list)\n",
    "                edge_properties_dict[\"cooccurring_domains_in_\"+domainoccurrence.name+\"_count\"]=domainoccurrence_len\n",
    "                edge_properties_dict[\"cooccurring_domains_in_\"+domainoccurrence.name+\"_boolean\"]=domainoccurrence_len>0\n",
    "                edge_properties_dict[\"human_readable_string_i\"]+=\" (\"+str(domainoccurrence_len)+\")\"\n",
    "                edge_properties_dict[\"human_readable_string_j\"]+=\",\\n\"+str(domainoccurrence_len)+\" in \"+domainoccurrence.name\n",
    "                                     \n",
    "            \n",
    "            \n",
    "            return edge_properties_dict\n",
    "\n",
    "\n",
    "        #start of the meat of the function\n",
    "        coocurrence_network=networkx.DiGraph() if generate_directed_graph else networkx.Graph()\n",
    "\n",
    "        \n",
    "        #draw nodes\n",
    "        if verbose:\n",
    "            print(\"drawing nodes...\")\n",
    "        for domain_accession in self.domain_to_protein_list_dict:\n",
    "            domain_tally=self.get_domain_protein_tally(domain_accession)\n",
    "            if(draw_node_when(domain_tally)):\n",
    "                name_string=pfam_id_name_description_df[\"Name\"].get(domain_accession, default=\"unknown_name\")\n",
    "\n",
    "                domain_in_mibig_list=mibig_domainoccurrences.get_domain_protein_list(domain_accession)\n",
    "                domain_in_uniprot_curated_catalytic_list=uniprot_curated_catalytic_domainoccurrences.get_domain_protein_list(domain_accession)\n",
    "                domain_in_brenda_list=brenda_domainoccurrences.get_domain_protein_list(domain_accession)\n",
    "                domain_in_mibig_list_with_accessions=[k+\":\"+v for k,v in uniprot_to_mibig_mapping_df.loc[domain_in_mibig_list,\"mibig_accession\"].iteritems()]\n",
    "                domain_in_brenda_list_with_accessions=[k+\":\"+v for k,v in uniprot_to_brenda_mapping_df.loc[domain_in_brenda_list,\"brenda_ec_number\"].iteritems()]\n",
    "                domain_in_uniprot_curated_catalytic_list_with_rhea_ids=[k+\":\"+v for k,v in uniprot_curated_catalytic_df.loc[domain_in_uniprot_curated_catalytic_list,\"Rhea Ids\"].iteritems()]\n",
    "\n",
    "\n",
    "                in_databases_abbreviation_list=[]\n",
    "                if len(domain_in_uniprot_curated_catalytic_list)>0:\n",
    "                    in_databases_abbreviation_list.append(\"U\")\n",
    "                if len(domain_in_mibig_list)>0:\n",
    "                    in_databases_abbreviation_list.append(\"M\")\n",
    "                if len(domain_in_brenda_list)>0:\n",
    "                    in_databases_abbreviation_list.append(\"B\")\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "                node_properties_dict={\n",
    "                        \"name\":name_string,\n",
    "                        \"description\":pfam_id_name_description_df[\"Description\"].get(domain_accession, default=\"unknown_description\"),\n",
    "                        \"long_description\":pfam_id_name_description_df[\"Long_description\"].get(domain_accession, default=\"unknown_long_description\"),\n",
    "                        \"annotation\":pfam_id_name_description_df[\"enzyme_annotation\"].get(domain_accession, default=\"unknown_annotation\"),\n",
    "                        \"occurrences_count\":domain_tally,\n",
    "\n",
    "                        \"human_readable_string_1\":name_string,\n",
    "                        \"human_readable_string_2\":name_string+\"\\n(\"+str(domain_tally)+(\" proteins)\" if domain_tally>1 else \" protein)\"),\n",
    "                        \"human_readable_string_3\":name_string+\"\\n\"+str(domain_tally)+(\" proteins\" if domain_tally>1 else \" protein\"),\n",
    "                        \"human_readable_string_4\":name_string+\"\\n(\"+str(domain_tally)+\")\",\n",
    "\n",
    "                        \"domain_in_uniprot_curated_catalytic_list\":truncate_list(domain_in_uniprot_curated_catalytic_list),\n",
    "                        \"domain_in_uniprot_curated_catalytic_count\":len(domain_in_uniprot_curated_catalytic_list),\n",
    "                        \"domain_in_uniprot_curated_catalytic_boolean\":len(domain_in_uniprot_curated_catalytic_list)>0,\n",
    "\n",
    "                        \"domain_in_mibig_list\":truncate_list(domain_in_mibig_list),\n",
    "                        \"domain_in_mibig_count\":len(domain_in_mibig_list),\n",
    "                        \"domain_in_mibig_boolean\":len(domain_in_mibig_list)>0,\n",
    "\n",
    "                        \"in_databases_abbreviations\":\"\".join(in_databases_abbreviation_list),\n",
    "                        }\n",
    "\n",
    "                for domainoccurrence in secondary_domaincooccurrences:\n",
    "                    domainoccurrence_list=domainoccurrence.get_domain_protein_list(domain_accession) #!\n",
    "                    domainoccurrence_len=len(domainoccurrence_list)\n",
    "                    assert \"occurrence_in_\"+domainoccurrence.name+\"_list\" not in node_properties_dict, \"cannot use duplicate names for secondary DomainOccurrences\"\n",
    "                    node_properties_dict[\"occurrence_in_\"+domainoccurrence.name+\"_list\"]=truncate_list(domainoccurrence_list)\n",
    "                    node_properties_dict[\"occurrence_in_\"+domainoccurrence.name+\"_count\"]=domainoccurrence_len\n",
    "                    node_properties_dict[\"occurrence_in_\"+domainoccurrence.name+\"_boolean\"]=domainoccurrence_len>0\n",
    "                    node_properties_dict[\"human_readable_string_3\"]+=\",\\n\"+str(domainoccurrence_len)+\" in \"+domainoccurrence.name\n",
    "                    node_properties_dict[\"human_readable_string_4\"]+=\" (\"+str(domainoccurrence_len)+\")\"\n",
    "                            \n",
    "                            \n",
    "                coocurrence_network.add_node(domain_accession,**node_properties_dict)\n",
    "\n",
    "        #draw edges\n",
    "\n",
    "        if verbose:\n",
    "            print(\"drawing edges...\")\n",
    "\n",
    "        if generate_directed_graph: #draw directed edges\n",
    "            count=0\n",
    "            for entry in self.domain_to_protein_list_dict:\n",
    "                if(count%500==0 and count>0 and verbose):\n",
    "                    print(\"drawing edges from node # {} out of {}\".format(count,len(self.domain_to_protein_list_dict)))\n",
    "                count+=1\n",
    "                domain_tally=self.get_domain_protein_tally(entry)\n",
    "                for other_entry in self.domain_to_protein_list_dict:\n",
    "                    if entry!=other_entry:\n",
    "                        entries_frozenset=frozenset([entry,other_entry])\n",
    "                        accessions_list=self.get_domain_pair_protein_list(entries_frozenset) #!\n",
    "                        tally=len(accessions_list)\n",
    "                        percent_coocurrence=round(tally/domain_tally*100,5)\n",
    "                        if draw_edge_when(tally,percent_coocurrence):\n",
    "\n",
    "                            edge_properties_dict=generate_edge_properties(accessions_list,entries_frozenset)\n",
    "                            in_databases_abbreviation_list=edge_properties_dict[\"cooccurring_domains_in_databases_abbreviation_list\"]\n",
    "                            edge_properties_dict[\"percent_cooccurrence\"]=percent_coocurrence\n",
    "                            edge_properties_dict[\"human_readable_string_1\"]=str(tally)+\"/\"+str(domain_tally)+\"\\n(\"+str(round(tally/domain_tally*100,2))+\"%)\"\n",
    "                            edge_properties_dict[\"human_readable_string_2\"]=str(tally)+\"/\"+str(domain_tally)\n",
    "                            edge_properties_dict[\"human_readable_string_3\"]=str(tally)+\"/\"+str(domain_tally)+\"\\n(\"+str(round(tally/domain_tally*100,2))+\"%) \"+\"\".join([\"(\"+x+\")\" for x in in_databases_abbreviation_list])\n",
    "                            edge_properties_dict[\"human_readable_string_4\"]=str(tally)+\"/\"+str(domain_tally)+\" (\"+str(int(round(tally/domain_tally*100,0)))+\"%)\\n\"+\"\".join([\"(\"+x+\")\" for x in in_databases_abbreviation_list])\n",
    "                            edge_properties_dict[\"human_readable_string_5\"]=str(tally)+\"/\"+str(domain_tally)+\" (\"+str(round(tally/domain_tally*100,2))+\"%)\"\n",
    "                            edge_properties_dict[\"human_readable_string_6\"]=str(tally)+\"/\"+str(domain_tally)+\" (\"+str(round(tally/domain_tally*100,2))+\"%)\"\n",
    "                            edge_properties_dict[\"human_readable_string_7\"]=str(tally)+\" (\"+str(round(tally/domain_tally*100,2))+\"%)\"\n",
    "                            \n",
    "                            if len(secondary_domaincooccurrences)>0:\n",
    "                                edge_properties_dict[\"human_readable_string_6\"]+=\" in \"+self.name\n",
    "                            \n",
    "                            for domainoccurrence in secondary_domaincooccurrences:\n",
    "                                domainoccurrence_list=domainoccurrence.get_domain_pair_protein_list(entries_frozenset)\n",
    "                                domainoccurrence_len=len(domainoccurrence_list)\n",
    "                                secondary_from_tally=domainoccurrence.get_domain_protein_tally(entry) #!\n",
    "                                if secondary_from_tally>0:\n",
    "                                    secondary_percent_cooccurrence=round(domainoccurrence_len/secondary_from_tally*100,5)\n",
    "                                    edge_properties_dict[\"percent_cooccurrence_in_\"+domainoccurrence.name]=secondary_percent_cooccurrence\n",
    "                                    edge_properties_dict[\"human_readable_string_5\"]+=\",\\n\"+str(domainoccurrence_len)+\"/\"+str(secondary_from_tally)+\" (\"+str(round(secondary_percent_cooccurrence,2))+\"%)\"\n",
    "                                    edge_properties_dict[\"human_readable_string_6\"]+=\",\\n\"+str(domainoccurrence_len)+\"/\"+str(secondary_from_tally)+\" (\"+str(round(secondary_percent_cooccurrence,2))+\"%)\"+\" in \"+domainoccurrence.name\n",
    "                                    edge_properties_dict[\"human_readable_string_7\"]+=\",\\n\"+str(domainoccurrence_len)+\" (\"+str(round(secondary_percent_cooccurrence,2))+\"%)\"\n",
    "                                else:\n",
    "                                    edge_properties_dict[\"percent_cooccurrence_in_\"+domainoccurrence.name]=\"N/A\"\n",
    "                                    edge_properties_dict[\"human_readable_string_5\"]+=\",\\n\"+str(domainoccurrence_len)+\"/\"+str(secondary_from_tally)\n",
    "                                    edge_properties_dict[\"human_readable_string_6\"]+=\",\\n\"+str(domainoccurrence_len)+\"/\"+str(secondary_from_tally)+\" in \"+domainoccurrence.name\n",
    "                                    edge_properties_dict[\"human_readable_string_7\"]+=\",\\n\"+str(domainoccurrence_len)\n",
    "\n",
    "                            \n",
    "                            coocurrence_network.add_edge(entry,other_entry, **edge_properties_dict)\n",
    "\n",
    "        else: #draw undirected edges\n",
    "            count=0\n",
    "            for domain_pair_frozenset, accessions_set in self.domain_pair_frozenset_to_protein_list_dict.items():\n",
    "                if(count%1000==0 and count>0 and verbose):\n",
    "                    print(\"drawing edge # {} out of {}\".format(count,len(domain_pair_to_protein_list_dict)))\n",
    "                tally=len(accessions_set)\n",
    "                if draw_edge_when(tally,100.0):\n",
    "                    edge_properties_dict=generate_edge_properties(list(accessions_set),domain_pair_frozenset)\n",
    "                    coocurrence_network.add_edge(*domain_pair_frozenset, **edge_properties_dict)\n",
    "\n",
    "\n",
    "\n",
    "        if verbose:\n",
    "            print(\"writing file \"+output_filename+\"...\")\n",
    "        with open(output_filename, 'w') as out_file:\n",
    "            json.dump(networkx.cytoscape_data(coocurrence_network), out_file)\n",
    "        if verbose:\n",
    "            print(\"completed writing file\")\n",
    "    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generate enzyme domain co-occurrence networks from PfamScan output (Fig. 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Processing curated Uniprot catalytic proteins\n",
      "now processing row 10000 of 225047\n",
      "now processing row 20000 of 225047\n",
      "now processing row 30000 of 225047\n",
      "now processing row 40000 of 225047\n",
      "now processing row 50000 of 225047\n",
      "now processing row 60000 of 225047\n",
      "now processing row 70000 of 225047\n",
      "now processing row 80000 of 225047\n",
      "now processing row 90000 of 225047\n",
      "now processing row 100000 of 225047\n",
      "now processing row 110000 of 225047\n",
      "now processing row 120000 of 225047\n",
      "now processing row 130000 of 225047\n",
      "now processing row 140000 of 225047\n",
      "now processing row 150000 of 225047\n",
      "now processing row 160000 of 225047\n",
      "now processing row 170000 of 225047\n",
      "now processing row 180000 of 225047\n",
      "now processing row 190000 of 225047\n",
      "now processing row 200000 of 225047\n",
      "now processing row 210000 of 225047\n",
      "now processing row 220000 of 225047\n",
      "Completed curated Uniprot catalytic proteins\n",
      "Processing Brenda proteins...\n",
      "now processing row 10000 of 31955\n",
      "now processing row 20000 of 31955\n",
      "now processing row 30000 of 31955\n",
      "Completed Brenda proteins\n",
      "Processing MIBiG proteins...\n",
      "now processing row 10000 of 30292\n",
      "now processing row 20000 of 30292\n",
      "now processing row 30000 of 30292\n",
      "Completed MIBiG proteins...\n",
      "Completed.\n"
     ]
    }
   ],
   "source": [
    "###Here we pre-process the uniprot entries annotated in various databases for easy indexing in a dictionary\n",
    "\n",
    "#General settings and filenames for this section.\n",
    "#We generated these files in the preparing_data notebook.\n",
    "uniprot_curated_catalytic_filename=\"Uniprot_curated_catalytic_activity.tsv\"\n",
    "uniprot_pathway_filename=\"uniprot_pathway_complete.tsv\"\n",
    "uniprot_to_mibig_mapping_filename=\"uniprot_to_mibig_merged_on_protein_id.tsv\"\n",
    "uniprot_to_brenda_mapping_filename=\"uniprot_to_brenda_merged.tsv\"\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "uniprot_curated_catalytic_df=pandas.read_table(uniprot_curated_catalytic_filename,index_col=0,dtype=\"str\",keep_default_na=False, na_values=[])\n",
    "uniprot_pathway_df=pandas.read_table(uniprot_pathway_filename,index_col=0,dtype=\"str\",keep_default_na=False, na_values=[])\n",
    "uniprot_to_mibig_mapping_df=pandas.read_table(uniprot_to_mibig_mapping_filename,index_col=0,dtype=\"str\",keep_default_na=False, na_values=[])\n",
    "uniprot_to_brenda_mapping_df=pandas.read_table(uniprot_to_brenda_mapping_filename,index_col=0,dtype=\"str\",keep_default_na=False, na_values=[])\n",
    "\n",
    "#combining uniprot curated catalytic and uniprot pathway\n",
    "uniprot_pathway_df.drop_duplicates(inplace=True)\n",
    "uniprot_curated_catalytic_df=pandas.concat(\n",
    "                [\n",
    "                uniprot_curated_catalytic_df,\n",
    "                uniprot_pathway_df.reindex(uniprot_pathway_df.index.difference(uniprot_curated_catalytic_df.index))\n",
    "                ],\n",
    "                verify_integrity=True).fillna(\"unknown\")\n",
    "\n",
    "print(\"Processing curated Uniprot catalytic proteins\")\n",
    "uniprot_curated_catalytic_domainoccurrences=DomainOccurrences(\"uniprot_curated_catalytic\",*preprocess_domain_pairs_in_uniprot_dataframe(uniprot_curated_catalytic_df))\n",
    "print(\"Completed curated Uniprot catalytic proteins\")\n",
    "\n",
    "print(\"Processing Brenda proteins...\")\n",
    "brenda_domainoccurrences=DomainOccurrences(\"brenda\",*preprocess_domain_pairs_in_uniprot_dataframe(uniprot_to_brenda_mapping_df))\n",
    "print(\"Completed Brenda proteins\")\n",
    "\n",
    "print(\"Processing MIBiG proteins...\")\n",
    "mibig_domainoccurrences=DomainOccurrences(\"mibig\",*preprocess_domain_pairs_in_uniprot_dataframe(uniprot_to_mibig_mapping_df))\n",
    "print(\"Completed MIBiG proteins...\")\n",
    "\n",
    "print(\"Completed.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "First we run PfamScan on fasta files of organisms' proteomes (.faa files) obtained from NCBI"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%bash\n",
    "export PERL5LIB=/mnt/d/domain_analysis/PfamScan:$PERL5LIB #change to wherever you have installed PfamScan\n",
    "./PfamScan/pfam_scan.pl -fasta GCF_000005845.2_ASM584v2_protein.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Ecoli_K12_MG1655.PfamScan -cpu 7\n",
    "./PfamScan/pfam_scan.pl -fasta GCF_000238295.2_Prub_2.0_protein.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Prubra_DSM6842.PfamScan -cpu 7\n",
    "./PfamScan/pfam_scan.pl -fasta GCA_000203835.1_ASM20383v1_protein.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Scoelicolor_A32-protein.fasta -cpu 7 &\n",
    "./PfamScan/pfam_scan.pl -fasta GCA_000016425.1_ASM1642v1_protein.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Stropica_CNB-440.PfamScan -cpu 7 &\n",
    "./PfamScan/pfam_scan.pl -fasta GCA_000012265.1_ASM1226v1_protein.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Pfluorescens_Pf5.PfamScan -cpu 7 &\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Then we run these python functions to convert that into a network that Cytoscape can read"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "processing Ecoli_K12_MG1655.PfamScan\n",
      "loading Ecoli_K12_MG1655.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 1597\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 702\n",
      "writing file Ecoli_K12_MG1655.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Ecoli_K12_MG1655.undirected_network.json...\n",
      "completed writing file\n",
      "processing Stropica_CNB-440.PfamScan\n",
      "loading Stropica_CNB-440.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 1844\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 644\n",
      "writing file Stropica_CNB-440.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Stropica_CNB-440.undirected_network.json...\n",
      "completed writing file\n",
      "processing Prubra_DSM6842.PfamScan\n",
      "loading Prubra_DSM6842.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 1700\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 675\n",
      "writing file Prubra_DSM6842.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Prubra_DSM6842.undirected_network.json...\n",
      "completed writing file\n",
      "processing Scoelicolor_A32.PfamScan\n",
      "loading Scoelicolor_A32.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 2938\n",
      "now processing protein #2000 out of 2938\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 775\n",
      "writing file Scoelicolor_A32.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Scoelicolor_A32.undirected_network.json...\n",
      "completed writing file\n",
      "processing Pfluorescens_Pf5.PfamScan\n",
      "loading Pfluorescens_Pf5.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 2264\n",
      "now processing protein #2000 out of 2264\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 739\n",
      "writing file Pfluorescens_Pf5.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Pfluorescens_Pf5.undirected_network.json...\n",
      "completed writing file\n",
      "Completed.\n"
     ]
    }
   ],
   "source": [
    "file_prefixes=[\"Ecoli_K12_MG1655\",\n",
    "               \"Stropica_CNB-440\",\n",
    "               \"Prubra_DSM6842\",\n",
    "               \"Scoelicolor_A32\",\n",
    "               \"Pfluorescens_Pf5\",\n",
    "              ]\n",
    "\n",
    "input_file_suffix=\".PfamScan\"\n",
    "output_file_suffix=\".json\"\n",
    "\n",
    "\n",
    "for prefix in file_prefixes:\n",
    "    pfamscan_filename=prefix+input_file_suffix\n",
    "    print(\"processing \"+pfamscan_filename)\n",
    "    output_network_filename=prefix+output_file_suffix\n",
    "    domainoccurrences=DomainOccurrences(prefix,*load_PfamScan_domain_pairs_we_care_about(pfamscan_filename))\n",
    "    domainoccurrences.generate_coocurrence_network(prefix+\".directed_network.json\")\n",
    "    domainoccurrences.generate_coocurrence_network(prefix+\".undirected_network.json\",generate_directed_graph=False)\n",
    "print(\"Completed.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Calculating percentage of multi-domain enzymes in various species (Fig. S1)\n",
    "You ought to fill in your own e-mail address and api-key for NCBI:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "email_for_ncbi=\"your@email.com\"\n",
    "ncbi_api_key=\"get_this_from_your_NCBI_account\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "requesting 194 proteins for Scoelicolor_A32\n",
      "response in 1.015055 s\n",
      "file Scoelicolor_A32_taxid_100226_entries_in_mibig_only.faa written\n",
      "requesting 182 proteins for Stropica_CNB-440\n",
      "response in 0.540926 s\n",
      "file Stropica_CNB-440_taxid_369723_entries_in_mibig_only.faa written\n",
      "requesting 144 proteins for Pfluorescens_Pf5\n",
      "response in 0.639825 s\n",
      "file Pfluorescens_Pf5_taxid_220664_entries_in_mibig_only.faa written\n",
      "Completed\n"
     ]
    }
   ],
   "source": [
    "import requests\n",
    "import time\n",
    "\n",
    "###general settings and filenames for this section\n",
    "mibig_to_ncbi_mapping_filename=\"ncbi_accessions_to_mibig.tsv\" #we generated this file in the preparing_data notebook\n",
    "\n",
    "\n",
    "#tuples of species and the taxid with which the BGCs are annotated in MIBiG\n",
    "species_of_interest=[(\"Scoelicolor_A32\",100226),\n",
    "                     (\"Stropica_CNB-440\",369723),\n",
    "                     (\"Pfluorescens_Pf5\",220664)]\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "mibig_to_ncbi_mapping_df=pandas.read_table(mibig_to_ncbi_mapping_filename)\n",
    "session = requests.Session()\n",
    "session.headers.update({'User-Agent': 'programmatic access by '+email_for_ncbi,\n",
    "                       'From':email_for_ncbi,\n",
    "                       })\n",
    "ncbi_efetch_base_url = \"https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?api_key=\"+ncbi_api_key+\"&email=\"+email_for_ncbi+\"&db=protein&rettype=fasta&id=\"\n",
    "\n",
    "for species_name,taxid in species_of_interest:\n",
    "    \n",
    "    output_filename=\"{}_taxid_{}_entries_in_mibig_only.faa\".format(species_name,taxid)\n",
    "    \n",
    "    ids_frozenset=frozenset(mibig_to_ncbi_mapping_df[mibig_to_ncbi_mapping_df[\"mibig_taxid\"]==taxid][\"ncbi_protein_accession\"].to_list())\n",
    "    request_url=ncbi_efetch_base_url+\",\".join(ids_frozenset)\n",
    "    print(\"requesting {} proteins for {}\".format(len(ids_frozenset),species_name))\n",
    "    r=session.get(request_url)\n",
    "    if(r.ok):\n",
    "        print(\"response in \"+str(r.elapsed.total_seconds())+\" s\")\n",
    "        if(r.text):\n",
    "            with open(output_filename,\"wt\") as file:\n",
    "                file.write(r.text)\n",
    "            print(\"file {} written\".format(output_filename))\n",
    "        else:\n",
    "            print(\"no results\")\n",
    "    else:\n",
    "        print(\"error: \"+str(r.status_code))\n",
    "    time.sleep(5)\n",
    "\n",
    "print(\"Completed\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "now we run these files through pfam_scan.pl, as well as the full proteomes for these same organisms."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%bash\n",
    "export PERL5LIB=/mnt/d/domain_analysis/PfamScan:$PERL5LIB #change to wherever you have installed PfamScan\n",
    "#mibig-only files\n",
    "./PfamScan/pfam_scan.pl -fasta Scoelicolor_A32_taxid_100226_entries_in_mibig_only.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Scoelicolor_A32_entries_in_mibig_only.PfamScan -cpu 7 &\n",
    "./PfamScan/pfam_scan.pl -fasta Stropica_CNB-440_taxid_369723_entries_in_mibig_only.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Stropica_CNB-440_entries_in_mibig_only.PfamScan -cpu 7\n",
    "./PfamScan/pfam_scan.pl -fasta Pfluorescens_Pf5_taxid_220664_entries_in_mibig_only.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Pfluorescens_Pf5_entries_in_mibig_only.PfamScan -cpu 7\n",
    "#full proteomes (these were already done in the previous section of the notebook, so you can skip them)\n",
    "./PfamScan/pfam_scan.pl -fasta GCA_000203835.1_ASM20383v1_protein.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Scoelicolor_A32-protein.fasta -cpu 7 &\n",
    "./PfamScan/pfam_scan.pl -fasta GCA_000016425.1_ASM1642v1_protein.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Stropica_CNB-440.PfamScan -cpu 7 &\n",
    "./PfamScan/pfam_scan.pl -fasta GCA_000012265.1_ASM1226v1_protein.faa -dir PfamScan -e_seq 0.001 -e_dom 0.001 -outfile Pfluorescens_Pf5.PfamScan -cpu 7 &"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "analyzing Scoelicolor_A32.PfamScan\n",
      "loading Scoelicolor_A32.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 2938\n",
      "now processing protein #2000 out of 2938\n",
      "analyzing Scoelicolor_A32_entries_in_mibig_only.PfamScan\n",
      "loading Scoelicolor_A32_entries_in_mibig_only.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "analyzing Stropica_CNB-440.PfamScan\n",
      "loading Stropica_CNB-440.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 1844\n",
      "analyzing Stropica_CNB-440_entries_in_mibig_only.PfamScan\n",
      "loading Stropica_CNB-440_entries_in_mibig_only.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "analyzing Pfluorescens_Pf5.PfamScan\n",
      "loading Pfluorescens_Pf5.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 2264\n",
      "now processing protein #2000 out of 2264\n",
      "analyzing Pfluorescens_Pf5_entries_in_mibig_only.PfamScan\n",
      "loading Pfluorescens_Pf5_entries_in_mibig_only.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "Completed.\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>proportion_multidomain</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>Scoelicolor_A32.PfamScan</th>\n",
       "      <td>0.022124</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Scoelicolor_A32_entries_in_mibig_only.PfamScan</th>\n",
       "      <td>0.100000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Stropica_CNB-440.PfamScan</th>\n",
       "      <td>0.033080</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Stropica_CNB-440_entries_in_mibig_only.PfamScan</th>\n",
       "      <td>0.091837</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Pfluorescens_Pf5.PfamScan</th>\n",
       "      <td>0.025618</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Pfluorescens_Pf5_entries_in_mibig_only.PfamScan</th>\n",
       "      <td>0.161765</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 proportion_multidomain\n",
       "Scoelicolor_A32.PfamScan                                       0.022124\n",
       "Scoelicolor_A32_entries_in_mibig_only.PfamScan                 0.100000\n",
       "Stropica_CNB-440.PfamScan                                      0.033080\n",
       "Stropica_CNB-440_entries_in_mibig_only.PfamScan                0.091837\n",
       "Pfluorescens_Pf5.PfamScan                                      0.025618\n",
       "Pfluorescens_Pf5_entries_in_mibig_only.PfamScan                0.161765"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "files_to_analyze=[\"Scoelicolor_A32.PfamScan\",\n",
    "                  \"Scoelicolor_A32_entries_in_mibig_only.PfamScan\",\n",
    "                  \"Stropica_CNB-440.PfamScan\",\n",
    "                  \"Stropica_CNB-440_entries_in_mibig_only.PfamScan\",\n",
    "                  \"Pfluorescens_Pf5.PfamScan\",\n",
    "                  \"Pfluorescens_Pf5_entries_in_mibig_only.PfamScan\"]\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "\n",
    "multidomain_series=pandas.Series(index=files_to_analyze,dtype='float')\n",
    "domain_count_df=pandas.DataFrame(columns=files_to_analyze)\n",
    "for filename in files_to_analyze:\n",
    "    print(\"analyzing \"+filename)\n",
    "    filtered_results_df=load_PfanScan_results_only_domains_we_care_about_dataframe(filename)\n",
    "    accession_to_number_of_domains_series=filtered_results_df[\"protein_id\"].value_counts()\n",
    "    proportion_multidomain=sum(accession_to_number_of_domains_series>1)/len(accession_to_number_of_domains_series)\n",
    "    multidomain_series[filename]=proportion_multidomain\n",
    "    domain_count_df[filename]=accession_to_number_of_domains_series.value_counts()\n",
    "    \n",
    "    \n",
    "print(\"Completed.\")\n",
    "pandas.DataFrame(multidomain_series,columns=[\"proportion_multidomain\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Scoelicolor_A32.PfamScan</th>\n",
       "      <th>Scoelicolor_A32_entries_in_mibig_only.PfamScan</th>\n",
       "      <th>Stropica_CNB-440.PfamScan</th>\n",
       "      <th>Stropica_CNB-440_entries_in_mibig_only.PfamScan</th>\n",
       "      <th>Pfluorescens_Pf5.PfamScan</th>\n",
       "      <th>Pfluorescens_Pf5_entries_in_mibig_only.PfamScan</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2873</td>\n",
       "      <td>99</td>\n",
       "      <td>1783</td>\n",
       "      <td>89</td>\n",
       "      <td>2206</td>\n",
       "      <td>57</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>48</td>\n",
       "      <td>5</td>\n",
       "      <td>48</td>\n",
       "      <td>1</td>\n",
       "      <td>48</td>\n",
       "      <td>8</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>5</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>7</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>9</td>\n",
       "      <td>5</td>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>2</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Scoelicolor_A32.PfamScan  Scoelicolor_A32_entries_in_mibig_only.PfamScan  \\\n",
       "1                      2873                                              99   \n",
       "2                        48                                               5   \n",
       "3                         5                                               1   \n",
       "4                         9                                               5   \n",
       "5                         3                                               0   \n",
       "\n",
       "   Stropica_CNB-440.PfamScan  Stropica_CNB-440_entries_in_mibig_only.PfamScan  \\\n",
       "1                       1783                                               89   \n",
       "2                         48                                                1   \n",
       "3                          5                                                3   \n",
       "4                          5                                                3   \n",
       "5                          3                                                2   \n",
       "\n",
       "   Pfluorescens_Pf5.PfamScan  Pfluorescens_Pf5_entries_in_mibig_only.PfamScan  \n",
       "1                       2206                                               57  \n",
       "2                         48                                                8  \n",
       "3                          7                                                3  \n",
       "4                          3                                                0  \n",
       "5                          0                                                0  "
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#if you wanted to see a more detailed distribution of the number of domains in enzymes...\n",
    "domain_count_df.fillna(0).astype('int').sort_index()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Generate enzyme domain co-occurrence network for all of uniprot (Figs. S2, S3, S14-17)\n",
    "Warning: this section requires a fair bit of RAM. Memory usage is roughly proportional to the number of domains included in the analysis."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "reading files...\n",
      "now on file #100 of 1745 (5.73%)\n",
      "now on file #200 of 1745 (11.46%)\n",
      "now on file #300 of 1745 (17.19%)\n",
      "now on file #400 of 1745 (22.92%)\n",
      "now on file #500 of 1745 (28.65%)\n",
      "now on file #600 of 1745 (34.38%)\n",
      "now on file #700 of 1745 (40.11%)\n",
      "now on file #800 of 1745 (45.85%)\n",
      "now on file #900 of 1745 (51.58%)\n",
      "now on file #1000 of 1745 (57.31%)\n",
      "now on file #1100 of 1745 (63.04%)\n",
      "now on file #1200 of 1745 (68.77%)\n",
      "now on file #1300 of 1745 (74.50%)\n",
      "now on file #1400 of 1745 (80.23%)\n",
      "now on file #1500 of 1745 (85.96%)\n",
      "now on file #1600 of 1745 (91.69%)\n",
      "now on file #1700 of 1745 (97.42%)\n",
      "finding pairs...\n",
      "analyzing domain pair #5000 out of 1521640 (0.33%) <PF00016(2),PF12681(1516)>\n",
      "analyzing domain pair #10000 out of 1521640 (0.66%) <PF00043(5),PF07726(1296)>\n",
      "analyzing domain pair #15000 out of 1521640 (0.99%) <PF00064(8),PF05138(1085)>\n",
      "analyzing domain pair #20000 out of 1521640 (1.31%) <PF00069(11),PF03571(883)>\n",
      "analyzing domain pair #25000 out of 1521640 (1.64%) <PF00074(14),PF02580(690)>\n",
      "analyzing domain pair #30000 out of 1521640 (1.97%) <PF00078(17),PF01764(506)>\n",
      "analyzing domain pair #35000 out of 1521640 (2.30%) <PF00089(20),PF01161(331)>\n",
      "analyzing domain pair #40000 out of 1521640 (2.63%) <PF00106(23),PF00549(165)>\n",
      "analyzing domain pair #45000 out of 1521640 (2.96%) <PF00109(25),PF18371(1726)>\n",
      "analyzing domain pair #50000 out of 1521640 (3.29%) <PF00115(28),PF13480(1575)>\n",
      "analyzing domain pair #55000 out of 1521640 (3.61%) <PF00120(31),PF09983(1433)>\n",
      "analyzing domain pair #60000 out of 1521640 (3.94%) <PF00128(34),PF07757(1300)>\n",
      "analyzing domain pair #65000 out of 1521640 (4.27%) <PF00136(37),PF06052(1176)>\n",
      "analyzing domain pair #70000 out of 1521640 (4.60%) <PF00145(40),PF04958(1061)>\n",
      "analyzing domain pair #75000 out of 1521640 (4.93%) <PF00150(43),PF04075(955)>\n",
      "analyzing domain pair #80000 out of 1521640 (5.26%) <PF00155(46),PF03364(858)>\n",
      "analyzing domain pair #85000 out of 1521640 (5.59%) <PF00162(49),PF02935(770)>\n",
      "analyzing domain pair #90000 out of 1521640 (5.91%) <PF00179(52),PF02581(691)>\n",
      "analyzing domain pair #95000 out of 1521640 (6.24%) <PF00186(55),PF02286(621)>\n",
      "analyzing domain pair #100000 out of 1521640 (6.57%) <PF00198(58),PF01983(560)>\n",
      "analyzing domain pair #105000 out of 1521640 (6.90%) <PF00202(61),PF01786(508)>\n",
      "analyzing domain pair #110000 out of 1521640 (7.23%) <PF00211(64),PF01613(465)>\n",
      "analyzing domain pair #115000 out of 1521640 (7.56%) <PF00218(67),PF01503(431)>\n",
      "analyzing domain pair #120000 out of 1521640 (7.89%) <PF00224(70),PF01398(406)>\n",
      "analyzing domain pair #125000 out of 1521640 (8.21%) <PF00233(73),PF01327(390)>\n",
      "analyzing domain pair #130000 out of 1521640 (8.54%) <PF00245(76),PF01288(383)>\n",
      "analyzing domain pair #135000 out of 1521640 (8.87%) <PF00251(79),PF01295(385)>\n",
      "analyzing domain pair #140000 out of 1521640 (9.20%) <PF00264(82),PF01351(396)>\n",
      "analyzing domain pair #145000 out of 1521640 (9.53%) <PF00268(85),PF01431(416)>\n",
      "analyzing domain pair #150000 out of 1521640 (9.86%) <PF00275(88),PF01532(445)>\n",
      "analyzing domain pair #155000 out of 1521640 (10.19%) <PF00288(91),PF01680(483)>\n",
      "analyzing domain pair #160000 out of 1521640 (10.51%) <PF00293(94),PF01887(530)>\n",
      "analyzing domain pair #165000 out of 1521640 (10.84%) <PF00300(97),PF02123(586)>\n",
      "analyzing domain pair #170000 out of 1521640 (11.17%) <PF00305(100),PF02426(651)>\n",
      "analyzing domain pair #175000 out of 1521640 (11.50%) <PF00326(103),PF02675(725)>\n",
      "analyzing domain pair #180000 out of 1521640 (11.83%) <PF00331(106),PF03099(808)>\n",
      "analyzing domain pair #185000 out of 1521640 (12.16%) <PF00342(109),PF03652(900)>\n",
      "analyzing domain pair #190000 out of 1521640 (12.49%) <PF00349(112),PF04358(1001)>\n",
      "analyzing domain pair #195000 out of 1521640 (12.82%) <PF00355(115),PF05362(1111)>\n",
      "analyzing domain pair #200000 out of 1521640 (13.14%) <PF00368(118),PF06560(1230)>\n",
      "analyzing domain pair #205000 out of 1521640 (13.47%) <PF00373(121),PF08704(1358)>\n",
      "analyzing domain pair #210000 out of 1521640 (13.80%) <PF00384(124),PF12146(1495)>\n",
      "analyzing domain pair #215000 out of 1521640 (14.13%) <PF00391(127),PF14606(1641)>\n",
      "analyzing domain pair #220000 out of 1521640 (14.46%) <PF00413(131),PF00632(183)>\n",
      "analyzing domain pair #225000 out of 1521640 (14.79%) <PF00423(134),PF01209(350)>\n",
      "analyzing domain pair #230000 out of 1521640 (15.12%) <PF00445(137),PF01870(526)>\n",
      "analyzing domain pair #235000 out of 1521640 (15.44%) <PF00457(140),PF02633(711)>\n",
      "analyzing domain pair #240000 out of 1521640 (15.77%) <PF00464(143),PF03706(905)>\n",
      "analyzing domain pair #245000 out of 1521640 (16.10%) <PF00478(146),PF05336(1108)>\n",
      "analyzing domain pair #250000 out of 1521640 (16.43%) <PF00487(149),PF07995(1320)>\n",
      "analyzing domain pair #255000 out of 1521640 (16.76%) <PF00491(152),PF12999(1541)>\n",
      "analyzing domain pair #260000 out of 1521640 (17.09%) <PF00507(156),PF00632(183)>\n",
      "analyzing domain pair #265000 out of 1521640 (17.42%) <PF00521(159),PF01474(425)>\n",
      "analyzing domain pair #270000 out of 1521640 (17.74%) <PF00544(162),PF02547(676)>\n",
      "analyzing domain pair #275000 out of 1521640 (18.07%) <PF00549(165),PF03916(936)>\n",
      "analyzing domain pair #280000 out of 1521640 (18.40%) <PF00565(168),PF06276(1205)>\n",
      "analyzing domain pair #285000 out of 1521640 (18.73%) <PF00578(171),PF11477(1483)>\n",
      "analyzing domain pair #290000 out of 1521640 (19.06%) <PF00587(175),PF00704(201)>\n",
      "analyzing domain pair #295000 out of 1521640 (19.39%) <PF00590(178),PF01750(500)>\n",
      "analyzing domain pair #300000 out of 1521640 (19.72%) <PF00623(181),PF03099(808)>\n",
      "analyzing domain pair #305000 out of 1521640 (20.04%) <PF00636(184),PF05543(1125)>\n",
      "analyzing domain pair #310000 out of 1521640 (20.37%) <PF00648(187),PF10364(1451)>\n",
      "analyzing domain pair #315000 out of 1521640 (20.70%) <PF00677(191),PF00796(233)>\n",
      "analyzing domain pair #320000 out of 1521640 (21.03%) <PF00692(194),PF02092(580)>\n",
      "analyzing domain pair #325000 out of 1521640 (21.36%) <PF00697(197),PF03916(936)>\n",
      "analyzing domain pair #330000 out of 1521640 (21.69%) <PF00703(200),PF07779(1301)>\n",
      "analyzing domain pair #335000 out of 1521640 (22.02%) <PF00709(203),PF16363(1675)>\n",
      "analyzing domain pair #340000 out of 1521640 (22.34%) <PF00719(207),PF01862(521)>\n",
      "analyzing domain pair #345000 out of 1521640 (22.67%) <PF00724(210),PF03756(916)>\n",
      "analyzing domain pair #350000 out of 1521640 (23.00%) <PF00731(213),PF07995(1320)>\n",
      "analyzing domain pair #355000 out of 1521640 (23.33%) <PF00743(216),PF18582(1733)>\n",
      "analyzing domain pair #360000 out of 1521640 (23.66%) <PF00759(220),PF02329(631)>\n",
      "analyzing domain pair #365000 out of 1521640 (23.99%) <PF00764(223),PF04989(1065)>\n",
      "analyzing domain pair #370000 out of 1521640 (24.32%) <PF00773(226),PF12388(1508)>\n",
      "analyzing domain pair #375000 out of 1521640 (24.64%) <PF00792(230),PF01536(446)>\n",
      "analyzing domain pair #380000 out of 1521640 (24.97%) <PF00796(233),PF03734(910)>\n",
      "analyzing domain pair #385000 out of 1521640 (25.30%) <PF00814(236),PF09171(1383)>\n",
      "analyzing domain pair #390000 out of 1521640 (25.63%) <PF00837(240),PF01231(361)>\n",
      "analyzing domain pair #395000 out of 1521640 (25.96%) <PF00850(243),PF03352(855)>\n",
      "analyzing domain pair #400000 out of 1521640 (26.29%) <PF00857(246),PF08704(1358)>\n",
      "analyzing domain pair #405000 out of 1521640 (26.62%) <PF00872(250),PF01265(376)>\n",
      "analyzing domain pair #410000 out of 1521640 (26.94%) <PF00883(253),PF03652(900)>\n",
      "analyzing domain pair #415000 out of 1521640 (27.27%) <PF00891(256),PF09983(1433)>\n",
      "analyzing domain pair #420000 out of 1521640 (27.60%) <PF00908(260),PF01729(491)>\n",
      "analyzing domain pair #425000 out of 1521640 (27.93%) <PF00920(263),PF04820(1045)>\n",
      "analyzing domain pair #430000 out of 1521640 (28.26%) <PF00926(266),PF13793(1608)>\n",
      "analyzing domain pair #435000 out of 1521640 (28.59%) <PF00941(270),PF02615(706)>\n",
      "analyzing domain pair #440000 out of 1521640 (28.92%) <PF00959(273),PF07669(1290)>\n",
      "analyzing domain pair #445000 out of 1521640 (29.24%) <PF00982(277),PF01431(416)>\n",
      "analyzing domain pair #450000 out of 1521640 (29.57%) <PF01011(280),PF04536(1021)>\n",
      "analyzing domain pair #455000 out of 1521640 (29.90%) <PF01026(283),PF14489(1635)>\n",
      "analyzing domain pair #460000 out of 1521640 (30.23%) <PF01040(287),PF03068(801)>\n",
      "analyzing domain pair #465000 out of 1521640 (30.56%) <PF01050(290),PF10100(1436)>\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "analyzing domain pair #470000 out of 1521640 (30.89%) <PF01059(294),PF02324(630)>\n",
      "analyzing domain pair #475000 out of 1521640 (31.22%) <PF01068(297),PF07519(1286)>\n",
      "analyzing domain pair #480000 out of 1521640 (31.54%) <PF01074(301),PF01786(508)>\n",
      "analyzing domain pair #485000 out of 1521640 (31.87%) <PF01081(304),PF06130(1185)>\n",
      "analyzing domain pair #490000 out of 1521640 (32.20%) <PF01095(308),PF01510(435)>\n",
      "analyzing domain pair #495000 out of 1521640 (32.53%) <PF01116(311),PF05652(1133)>\n",
      "analyzing domain pair #500000 out of 1521640 (32.86%) <PF01124(315),PF01411(411)>\n",
      "analyzing domain pair #505000 out of 1521640 (33.19%) <PF01128(318),PF05580(1130)>\n",
      "analyzing domain pair #510000 out of 1521640 (33.52%) <PF01137(322),PF01512(436)>\n",
      "analyzing domain pair #515000 out of 1521640 (33.85%) <PF01142(325),PF06052(1176)>\n",
      "analyzing domain pair #520000 out of 1521640 (34.17%) <PF01154(329),PF01791(510)>\n",
      "analyzing domain pair #525000 out of 1521640 (34.50%) <PF01168(332),PF07335(1271)>\n",
      "analyzing domain pair #530000 out of 1521640 (34.83%) <PF01175(336),PF02333(633)>\n",
      "analyzing domain pair #535000 out of 1521640 (35.16%) <PF01182(339),PF09520(1415)>\n",
      "analyzing domain pair #540000 out of 1521640 (35.49%) <PF01192(343),PF03079(805)>\n",
      "analyzing domain pair #545000 out of 1521640 (35.82%) <PF01204(346),PF13793(1608)>\n",
      "analyzing domain pair #550000 out of 1521640 (36.15%) <PF01209(350),PF04607(1026)>\n",
      "analyzing domain pair #555000 out of 1521640 (36.47%) <PF01219(354),PF01597(460)>\n",
      "analyzing domain pair #560000 out of 1521640 (36.80%) <PF01223(357),PF07726(1296)>\n",
      "analyzing domain pair #565000 out of 1521640 (37.13%) <PF01231(361),PF02878(758)>\n",
      "analyzing domain pair #570000 out of 1521640 (37.46%) <PF01234(364),PF14028(1615)>\n",
      "analyzing domain pair #575000 out of 1521640 (37.79%) <PF01243(368),PF05292(1105)>\n",
      "analyzing domain pair #580000 out of 1521640 (38.12%) <PF01259(372),PF02263(611)>\n",
      "analyzing domain pair #585000 out of 1521640 (38.45%) <PF01264(375),PF12241(1501)>\n",
      "analyzing domain pair #590000 out of 1521640 (38.77%) <PF01270(379),PF04723(1035)>\n",
      "analyzing domain pair #595000 out of 1521640 (39.10%) <PF01288(383),PF02122(585)>\n",
      "analyzing domain pair #600000 out of 1521640 (39.43%) <PF01301(386),PF12388(1508)>\n",
      "analyzing domain pair #605000 out of 1521640 (39.76%) <PF01327(390),PF05139(1086)>\n",
      "analyzing domain pair #610000 out of 1521640 (40.09%) <PF01339(394),PF02557(680)>\n",
      "analyzing domain pair #615000 out of 1521640 (40.42%) <PF01353(397),PF14501(1636)>\n",
      "analyzing domain pair #620000 out of 1521640 (40.75%) <PF01370(401),PF07156(1258)>\n",
      "analyzing domain pair #625000 out of 1521640 (41.07%) <PF01397(405),PF03632(896)>\n",
      "analyzing domain pair #630000 out of 1521640 (41.40%) <PF01406(409),PF01959(550)>\n",
      "analyzing domain pair #635000 out of 1521640 (41.73%) <PF01416(412),PF13304(1551)>\n",
      "analyzing domain pair #640000 out of 1521640 (42.06%) <PF01431(416),PF06699(1233)>\n",
      "analyzing domain pair #645000 out of 1521640 (42.39%) <PF01443(420),PF03881(931)>\n",
      "analyzing domain pair #650000 out of 1521640 (42.72%) <PF01470(424),PF02390(645)>\n",
      "analyzing domain pair #655000 out of 1521640 (43.05%) <PF01487(427),PF16868(1691)>\n",
      "analyzing domain pair #660000 out of 1521640 (43.37%) <PF01503(431),PF09983(1433)>\n",
      "analyzing domain pair #665000 out of 1521640 (43.70%) <PF01510(435),PF06180(1191)>\n",
      "analyzing domain pair #670000 out of 1521640 (44.03%) <PF01520(439),PF04127(965)>\n",
      "analyzing domain pair #675000 out of 1521640 (44.36%) <PF01529(443),PF02868(755)>\n",
      "analyzing domain pair #680000 out of 1521640 (44.69%) <PF01541(447),PF01993(561)>\n",
      "analyzing domain pair #685000 out of 1521640 (45.02%) <PF01553(450),PF16365(1676)>\n",
      "analyzing domain pair #690000 out of 1521640 (45.35%) <PF01569(454),PF12456(1510)>\n",
      "analyzing domain pair #695000 out of 1521640 (45.67%) <PF01593(458),PF08715(1360)>\n",
      "analyzing domain pair #700000 out of 1521640 (46.00%) <PF01609(462),PF06479(1226)>\n",
      "analyzing domain pair #705000 out of 1521640 (46.33%) <PF01625(466),PF05336(1108)>\n",
      "analyzing domain pair #710000 out of 1521640 (46.66%) <PF01640(470),PF04387(1006)>\n",
      "analyzing domain pair #715000 out of 1521640 (46.99%) <PF01645(474),PF03796(920)>\n",
      "analyzing domain pair #720000 out of 1521640 (47.32%) <PF01660(478),PF03328(850)>\n",
      "analyzing domain pair #725000 out of 1521640 (47.65%) <PF01678(482),PF03061(796)>\n",
      "analyzing domain pair #730000 out of 1521640 (47.97%) <PF01704(486),PF02878(758)>\n",
      "analyzing domain pair #735000 out of 1521640 (48.30%) <PF01728(490),PF02729(736)>\n",
      "analyzing domain pair #740000 out of 1521640 (48.63%) <PF01735(494),PF02685(730)>\n",
      "analyzing domain pair #745000 out of 1521640 (48.96%) <PF01746(498),PF02769(740)>\n",
      "analyzing domain pair #750000 out of 1521640 (49.29%) <PF01755(502),PF02910(766)>\n",
      "analyzing domain pair #755000 out of 1521640 (49.62%) <PF01764(506),PF03099(808)>\n",
      "analyzing domain pair #760000 out of 1521640 (49.95%) <PF01791(510),PF03415(866)>\n",
      "analyzing domain pair #765000 out of 1521640 (50.27%) <PF01804(514),PF03969(940)>\n",
      "analyzing domain pair #770000 out of 1521640 (50.60%) <PF01832(518),PF04655(1030)>\n",
      "analyzing domain pair #775000 out of 1521640 (50.93%) <PF01864(522),PF05681(1136)>\n",
      "analyzing domain pair #780000 out of 1521640 (51.26%) <PF01870(526),PF07156(1258)>\n",
      "analyzing domain pair #785000 out of 1521640 (51.59%) <PF01887(530),PF09409(1396)>\n",
      "analyzing domain pair #790000 out of 1521640 (51.92%) <PF01902(534),PF13302(1550)>\n",
      "analyzing domain pair #795000 out of 1521640 (52.25%) <PF01916(538),PF18238(1720)>\n",
      "analyzing domain pair #800000 out of 1521640 (52.57%) <PF01936(543),PF02614(705)>\n",
      "analyzing domain pair #805000 out of 1521640 (52.90%) <PF01950(547),PF03737(911)>\n",
      "analyzing domain pair #810000 out of 1521640 (53.23%) <PF01960(551),PF05652(1133)>\n",
      "analyzing domain pair #815000 out of 1521640 (53.56%) <PF01968(555),PF09009(1371)>\n",
      "analyzing domain pair #820000 out of 1521640 (53.89%) <PF01981(559),PF14360(1625)>\n",
      "analyzing domain pair #825000 out of 1521640 (54.22%) <PF02005(564),PF02649(715)>\n",
      "analyzing domain pair #830000 out of 1521640 (54.55%) <PF02015(568),PF04378(1005)>\n",
      "analyzing domain pair #835000 out of 1521640 (54.88%) <PF02056(572),PF07931(1311)>\n",
      "analyzing domain pair #840000 out of 1521640 (55.20%) <PF02075(576),PF14437(1633)>\n",
      "analyzing domain pair #845000 out of 1521640 (55.53%) <PF02102(581),PF03099(808)>\n",
      "analyzing domain pair #850000 out of 1521640 (55.86%) <PF02122(585),PF05935(1166)>\n",
      "analyzing domain pair #855000 out of 1521640 (56.19%) <PF02129(589),PF12942(1540)>\n",
      "analyzing domain pair #860000 out of 1521640 (56.52%) <PF02153(594),PF02971(780)>\n",
      "analyzing domain pair #865000 out of 1521640 (56.85%) <PF02195(598),PF06176(1190)>\n",
      "analyzing domain pair #870000 out of 1521640 (57.18%) <PF02223(602),PF14040(1616)>\n",
      "analyzing domain pair #875000 out of 1521640 (57.50%) <PF02241(607),PF03803(921)>\n",
      "analyzing domain pair #880000 out of 1521640 (57.83%) <PF02263(611),PF09171(1383)>\n",
      "analyzing domain pair #885000 out of 1521640 (58.16%) <PF02275(616),PF02700(733)>\n",
      "analyzing domain pair #890000 out of 1521640 (58.49%) <PF02285(620),PF06616(1231)>\n",
      "analyzing domain pair #895000 out of 1521640 (58.82%) <PF02297(625),PF02300(626)>\n",
      "analyzing domain pair #900000 out of 1521640 (59.15%) <PF02317(629),PF05881(1160)>\n",
      "analyzing domain pair #905000 out of 1521640 (59.48%) <PF02333(633),PF17990(1710)>\n",
      "analyzing domain pair #910000 out of 1521640 (59.80%) <PF02358(638),PF05996(1170)>\n",
      "analyzing domain pair #915000 out of 1521640 (60.13%) <PF02384(643),PF02435(655)>\n",
      "analyzing domain pair #920000 out of 1521640 (60.46%) <PF02395(647),PF07212(1261)>\n",
      "analyzing domain pair #925000 out of 1521640 (60.79%) <PF02431(652),PF03034(791)>\n",
      "analyzing domain pair #930000 out of 1521640 (61.12%) <PF02445(656),PF09983(1433)>\n",
      "analyzing domain pair #935000 out of 1521640 (61.45%) <PF02485(661),PF04406(1008)>\n",
      "analyzing domain pair #940000 out of 1521640 (61.78%) <PF02507(665),PF16715(1686)>\n",
      "analyzing domain pair #945000 out of 1521640 (62.10%) <PF02517(670),PF07858(1306)>\n",
      "analyzing domain pair #950000 out of 1521640 (62.43%) <PF02543(675),PF04041(951)>\n",
      "analyzing domain pair #955000 out of 1521640 (62.76%) <PF02552(679),PF16670(1685)>\n",
      "analyzing domain pair #960000 out of 1521640 (63.09%) <PF02568(684),PF09043(1375)>\n",
      "analyzing domain pair #965000 out of 1521640 (63.42%) <PF02574(689),PF05153(1090)>\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "analyzing domain pair #970000 out of 1521640 (63.75%) <PF02590(694),PF03200(830)>\n",
      "analyzing domain pair #975000 out of 1521640 (64.08%) <PF02601(698),PF14592(1640)>\n",
      "analyzing domain pair #980000 out of 1521640 (64.40%) <PF02611(703),PF09665(1425)>\n",
      "analyzing domain pair #985000 out of 1521640 (64.73%) <PF02621(708),PF06737(1235)>\n",
      "analyzing domain pair #990000 out of 1521640 (65.06%) <PF02635(713),PF05019(1070)>\n",
      "analyzing domain pair #995000 out of 1521640 (65.39%) <PF02657(718),PF03856(930)>\n",
      "analyzing domain pair #1000000 out of 1521640 (65.72%) <PF02668(723),PF03141(815)>\n",
      "analyzing domain pair #1005000 out of 1521640 (66.05%) <PF02678(727),PF18678(1741)>\n",
      "analyzing domain pair #1010000 out of 1521640 (66.38%) <PF02689(732),PF16123(1671)>\n",
      "analyzing domain pair #1015000 out of 1521640 (66.70%) <PF02732(737),PF14378(1626)>\n",
      "analyzing domain pair #1020000 out of 1521640 (67.03%) <PF02776(742),PF13759(1606)>\n",
      "analyzing domain pair #1025000 out of 1521640 (67.36%) <PF02786(747),PF13844(1611)>\n",
      "analyzing domain pair #1030000 out of 1521640 (67.69%) <PF02826(752),PF14606(1641)>\n",
      "analyzing domain pair #1035000 out of 1521640 (68.02%) <PF02875(757),PF17171(1696)>\n",
      "analyzing domain pair #1040000 out of 1521640 (68.35%) <PF02902(763),PF03060(795)>\n",
      "analyzing domain pair #1045000 out of 1521640 (68.68%) <PF02923(768),PF03706(905)>\n",
      "analyzing domain pair #1050000 out of 1521640 (69.00%) <PF02943(773),PF04762(1040)>\n",
      "analyzing domain pair #1055000 out of 1521640 (69.33%) <PF02963(778),PF06249(1200)>\n",
      "analyzing domain pair #1060000 out of 1521640 (69.66%) <PF02979(783),PF09195(1385)>\n",
      "analyzing domain pair #1065000 out of 1521640 (69.99%) <PF03022(788),PF13649(1595)>\n",
      "analyzing domain pair #1070000 out of 1521640 (70.32%) <PF03059(794),PF03512(880)>\n",
      "analyzing domain pair #1075000 out of 1521640 (70.65%) <PF03065(799),PF05724(1145)>\n",
      "analyzing domain pair #1080000 out of 1521640 (70.98%) <PF03074(804),PF10014(1435)>\n",
      "analyzing domain pair #1085000 out of 1521640 (71.30%) <PF03116(810),PF03150(816)>\n",
      "analyzing domain pair #1090000 out of 1521640 (71.63%) <PF03141(815),PF05889(1161)>\n",
      "analyzing domain pair #1095000 out of 1521640 (71.96%) <PF03162(820),PF12856(1531)>\n",
      "analyzing domain pair #1100000 out of 1521640 (72.29%) <PF03184(826),PF04406(1008)>\n",
      "analyzing domain pair #1105000 out of 1521640 (72.62%) <PF03201(831),PF09983(1433)>\n",
      "analyzing domain pair #1110000 out of 1521640 (72.95%) <PF03254(837),PF04204(976)>\n",
      "analyzing domain pair #1115000 out of 1521640 (73.28%) <PF03283(842),PF10459(1456)>\n",
      "analyzing domain pair #1120000 out of 1521640 (73.60%) <PF03313(848),PF04989(1065)>\n",
      "analyzing domain pair #1125000 out of 1521640 (73.93%) <PF03332(853),PF13701(1600)>\n",
      "analyzing domain pair #1130000 out of 1521640 (74.26%) <PF03372(859),PF07366(1275)>\n",
      "analyzing domain pair #1135000 out of 1521640 (74.59%) <PF03414(865),PF04252(986)>\n",
      "analyzing domain pair #1140000 out of 1521640 (74.92%) <PF03435(870),PF13759(1606)>\n",
      "analyzing domain pair #1145000 out of 1521640 (75.25%) <PF03492(876),PF09171(1383)>\n",
      "analyzing domain pair #1150000 out of 1521640 (75.58%) <PF03567(882),PF06230(1196)>\n",
      "analyzing domain pair #1155000 out of 1521640 (75.90%) <PF03587(888),PF04820(1045)>\n",
      "analyzing domain pair #1160000 out of 1521640 (76.23%) <PF03626(894),PF03856(930)>\n",
      "analyzing domain pair #1165000 out of 1521640 (76.56%) <PF03644(899),PF17101(1695)>\n",
      "analyzing domain pair #1170000 out of 1521640 (76.89%) <PF03706(905),PF14765(1646)>\n",
      "analyzing domain pair #1175000 out of 1521640 (77.22%) <PF03737(911),PF14437(1633)>\n",
      "analyzing domain pair #1180000 out of 1521640 (77.55%) <PF03767(917),PF15159(1656)>\n",
      "analyzing domain pair #1185000 out of 1521640 (77.88%) <PF03808(923),PF18130(1715)>\n",
      "analyzing domain pair #1190000 out of 1521640 (78.21%) <PF03856(930),PF04321(996)>\n",
      "analyzing domain pair #1195000 out of 1521640 (78.53%) <PF03916(936),PF05652(1133)>\n",
      "analyzing domain pair #1200000 out of 1521640 (78.86%) <PF03976(942),PF07858(1306)>\n",
      "analyzing domain pair #1205000 out of 1521640 (79.19%) <PF04029(948),PF12671(1515)>\n",
      "analyzing domain pair #1210000 out of 1521640 (79.52%) <PF04075(955),PF04188(971)>\n",
      "analyzing domain pair #1215000 out of 1521640 (79.85%) <PF04113(961),PF07156(1258)>\n",
      "analyzing domain pair #1220000 out of 1521640 (80.18%) <PF04140(967),PF13532(1581)>\n",
      "analyzing domain pair #1225000 out of 1521640 (80.51%) <PF04197(974),PF05996(1170)>\n",
      "analyzing domain pair #1230000 out of 1521640 (80.83%) <PF04226(980),PF13449(1571)>\n",
      "analyzing domain pair #1235000 out of 1521640 (81.16%) <PF04261(987),PF06983(1251)>\n",
      "analyzing domain pair #1240000 out of 1521640 (81.49%) <PF04277(993),PF18485(1730)>\n",
      "analyzing domain pair #1245000 out of 1521640 (81.82%) <PF04349(1000),PF12241(1501)>\n",
      "analyzing domain pair #1250000 out of 1521640 (82.15%) <PF04389(1007),PF08013(1321)>\n",
      "analyzing domain pair #1255000 out of 1521640 (82.48%) <PF04439(1014),PF06176(1190)>\n",
      "analyzing domain pair #1260000 out of 1521640 (82.81%) <PF04536(1021),PF05336(1108)>\n",
      "analyzing domain pair #1265000 out of 1521640 (83.13%) <PF04613(1028),PF05034(1075)>\n",
      "analyzing domain pair #1270000 out of 1521640 (83.46%) <PF04723(1035),PF05159(1091)>\n",
      "analyzing domain pair #1275000 out of 1521640 (83.79%) <PF04775(1042),PF05866(1156)>\n",
      "analyzing domain pair #1280000 out of 1521640 (84.12%) <PF04864(1049),PF07307(1270)>\n",
      "analyzing domain pair #1285000 out of 1521640 (84.45%) <PF04922(1056),PF09983(1433)>\n",
      "analyzing domain pair #1290000 out of 1521640 (84.78%) <PF04962(1063),PF14736(1645)>\n",
      "analyzing domain pair #1295000 out of 1521640 (85.11%) <PF05023(1071),PF06699(1233)>\n",
      "analyzing domain pair #1300000 out of 1521640 (85.43%) <PF05067(1078),PF13302(1550)>\n",
      "analyzing domain pair #1305000 out of 1521640 (85.76%) <PF05139(1086),PF07156(1258)>\n",
      "analyzing domain pair #1310000 out of 1521640 (86.09%) <PF05175(1093),PF16542(1680)>\n",
      "analyzing domain pair #1315000 out of 1521640 (86.42%) <PF05221(1101),PF12388(1508)>\n",
      "analyzing domain pair #1320000 out of 1521640 (86.75%) <PF05342(1109),PF09419(1400)>\n",
      "analyzing domain pair #1325000 out of 1521640 (87.08%) <PF05416(1117),PF08672(1356)>\n",
      "analyzing domain pair #1330000 out of 1521640 (87.41%) <PF05543(1125),PF09056(1376)>\n",
      "analyzing domain pair #1335000 out of 1521640 (87.73%) <PF05652(1133),PF10551(1460)>\n",
      "analyzing domain pair #1340000 out of 1521640 (88.06%) <PF05704(1141),PF13793(1608)>\n",
      "analyzing domain pair #1345000 out of 1521640 (88.39%) <PF05826(1150),PF06479(1226)>\n",
      "analyzing domain pair #1350000 out of 1521640 (88.72%) <PF05873(1158),PF12456(1510)>\n",
      "analyzing domain pair #1355000 out of 1521640 (89.05%) <PF05958(1167),PF07478(1281)>\n",
      "analyzing domain pair #1360000 out of 1521640 (89.38%) <PF06050(1175),PF17297(1701)>\n",
      "analyzing domain pair #1365000 out of 1521640 (89.71%) <PF06114(1184),PF14360(1625)>\n",
      "analyzing domain pair #1370000 out of 1521640 (90.03%) <PF06202(1193),PF14414(1630)>\n",
      "analyzing domain pair #1375000 out of 1521640 (90.36%) <PF06253(1202),PF18164(1716)>\n",
      "analyzing domain pair #1380000 out of 1521640 (90.69%) <PF06330(1212),PF08543(1351)>\n",
      "analyzing domain pair #1385000 out of 1521640 (91.02%) <PF06433(1221),PF13793(1608)>\n",
      "analyzing domain pair #1390000 out of 1521640 (91.35%) <PF06616(1231),PF09983(1433)>\n",
      "analyzing domain pair #1395000 out of 1521640 (91.68%) <PF06807(1241),PF08704(1358)>\n",
      "analyzing domain pair #1400000 out of 1521640 (92.01%) <PF06983(1251),PF09171(1383)>\n",
      "analyzing domain pair #1405000 out of 1521640 (92.33%) <PF07212(1261),PF12388(1508)>\n",
      "analyzing domain pair #1410000 out of 1521640 (92.66%) <PF07335(1271),PF18582(1733)>\n",
      "analyzing domain pair #1415000 out of 1521640 (92.99%) <PF07479(1282),PF13673(1596)>\n",
      "analyzing domain pair #1420000 out of 1521640 (93.32%) <PF07685(1293),PF13529(1580)>\n",
      "analyzing domain pair #1425000 out of 1521640 (93.65%) <PF07835(1304),PF16670(1685)>\n",
      "analyzing domain pair #1430000 out of 1521640 (93.98%) <PF07977(1316),PF11477(1483)>\n",
      "analyzing domain pair #1435000 out of 1521640 (94.31%) <PF08240(1328),PF09665(1425)>\n",
      "analyzing domain pair #1440000 out of 1521640 (94.63%) <PF08423(1340),PF12464(1511)>\n",
      "analyzing domain pair #1445000 out of 1521640 (94.96%) <PF08548(1352),PF18678(1741)>\n",
      "analyzing domain pair #1450000 out of 1521640 (95.29%) <PF08883(1365),PF18630(1736)>\n",
      "analyzing domain pair #1455000 out of 1521640 (95.62%) <PF09126(1379),PF12897(1535)>\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "analyzing domain pair #1460000 out of 1521640 (95.95%) <PF09293(1393),PF12852(1530)>\n",
      "analyzing domain pair #1465000 out of 1521640 (96.28%) <PF09491(1407),PF18257(1721)>\n",
      "analyzing domain pair #1470000 out of 1521640 (96.61%) <PF09631(1423),PF10662(1465)>\n",
      "analyzing domain pair #1475000 out of 1521640 (96.93%) <PF10128(1439),PF10662(1465)>\n",
      "analyzing domain pair #1480000 out of 1521640 (97.26%) <PF10432(1455),PF18257(1721)>\n",
      "analyzing domain pair #1485000 out of 1521640 (97.59%) <PF11266(1473),PF16867(1690)>\n",
      "analyzing domain pair #1490000 out of 1521640 (97.92%) <PF11991(1492),PF18566(1731)>\n",
      "analyzing domain pair #1495000 out of 1521640 (98.25%) <PF12578(1513),PF16116(1670)>\n",
      "analyzing domain pair #1500000 out of 1521640 (98.58%) <PF12899(1536),PF14437(1633)>\n",
      "analyzing domain pair #1505000 out of 1521640 (98.91%) <PF13378(1562),PF13486(1576)>\n",
      "analyzing domain pair #1510000 out of 1521640 (99.24%) <PF13614(1591),PF18582(1733)>\n",
      "analyzing domain pair #1515000 out of 1521640 (99.56%) <PF14412(1629),PF15785(1660)>\n",
      "analyzing domain pair #1520000 out of 1521640 (99.89%) <PF16822(1687),PF17297(1701)>\n",
      "creating DomainOccurrences object...\n",
      "Completed\n"
     ]
    }
   ],
   "source": [
    "#we generated the file in the folder below in the preparing_data notebook\n",
    "#each file, whose filename signifies a domain accession number, contains a list of uniprot protein accessions it is found in\n",
    "uniprot_domain_files_prefix=\"D:/domain_analysis/pfam_to_uniprot_files/\"\n",
    "\n",
    "\n",
    "\n",
    "domains_to_uniprot_frozensets_dict={}\n",
    "num_domains=len(domains_we_care_about)\n",
    "domain_pairs_to_uniprot_frozensets_dict={}\n",
    "print(\"reading files...\")\n",
    "for i,domain_accession in enumerate(domains_we_care_about):\n",
    "    if(i%100==0 and i>0):\n",
    "        print(\"now on file #{} of {} ({:.2f}%)\".format(i,num_domains,100.0*i/num_domains))\n",
    "    with open(uniprot_domain_files_prefix+domain_accession, \"tr\") as file:\n",
    "        domains_to_uniprot_frozensets_dict[domain_accession]=frozenset(file.read().splitlines())\n",
    "print(\"finding pairs...\")\n",
    "total_combinations=num_domains*(num_domains-1)//2\n",
    "counter=0\n",
    "for i in range(num_domains):\n",
    "    domain1_accession=domains_we_care_about[i]\n",
    "    domain1_uniprot_frozenset=domains_to_uniprot_frozensets_dict[domain1_accession]\n",
    "    for j in range(i+1,num_domains):\n",
    "        domain2_accession=domains_we_care_about[j]\n",
    "        if(counter%5000==0 and counter>0):\n",
    "            print(\"analyzing domain pair #{} out of {} ({:.2f}%) <{}({}),{}({})>\".format(counter,total_combinations,100.0*counter/total_combinations,domain1_accession,i,domain2_accession,j))\n",
    "        counter+=1\n",
    "        domain2_uniprot_frozenset=domains_to_uniprot_frozensets_dict[domain2_accession]\n",
    "        cooccurring_list=domain1_uniprot_frozenset.intersection(domain2_uniprot_frozenset)\n",
    "        domain_pairs_to_uniprot_frozensets_dict[frozenset([domain1_accession,domain2_accession])]=cooccurring_list\n",
    "print(\"creating DomainOccurrences object...\")        \n",
    "all_of_uniprot_domainoccurrences=DomainOccurrences(\"uniprot\",domains_to_uniprot_frozensets_dict,domain_pairs_to_uniprot_frozensets_dict)        \n",
    "print(\"Completed\")        \n",
    "#The time this cell takes scales by the number of domains analyzed squared.\n",
    "#For all domains annotated as \"y\",\"s\",\"e\", or \"m\" (1745 domains), it takes about 15 minutes on my computer. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 1745\n",
      "drawing edges from node # 1000 out of 1745\n",
      "drawing edges from node # 1500 out of 1745\n",
      "writing file all_of_uniprot-ysem-10_or_more.network.json...\n",
      "completed writing file\n"
     ]
    }
   ],
   "source": [
    "#draw the all-of-uniprot network.\n",
    "#change the lambdas passed as arguments below if you want stricter requirements for drawing an edge and/or node\n",
    "#(just make sure that when you draw an edge, the nodes exist)\n",
    "\n",
    "min_number_of_domain_pair_occurrences=10\n",
    "all_of_uniprot_domainoccurrences.generate_coocurrence_network(\n",
    "                      \"all_of_uniprot-{}-{}_or_more.network.json\".format(\n",
    "                          \"\".join(annotation_categories_we_care_about),\n",
    "                          min_number_of_domain_pair_occurrences),\n",
    "                      draw_node_when=(lambda occurrence: occurrence>=min_number_of_domain_pair_occurrences),\n",
    "                      draw_edge_when=(lambda absolute_coocurrence,pct_coocurrence: absolute_coocurrence>=min_number_of_domain_pair_occurrences),\n",
    "                      max_entries_in_list=500\n",
    "                     )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can also re-generate the single-genome networks but this time annotated with information derived from the all-of-uniprot cooccurrences.\n",
    "\n",
    "(this section re-uses the PfamScan results from the section for generating the network for fig. 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "processing Ecoli_K12_MG1655.PfamScan\n",
      "loading Ecoli_K12_MG1655.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 1597\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 702\n",
      "writing file Ecoli_K12_MG1655_and_uniprot.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Ecoli_K12_MG1655_and_uniprot.undirected_network.json...\n",
      "completed writing file\n",
      "processing Stropica_CNB-440.PfamScan\n",
      "loading Stropica_CNB-440.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 1844\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 644\n",
      "writing file Stropica_CNB-440_and_uniprot.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Stropica_CNB-440_and_uniprot.undirected_network.json...\n",
      "completed writing file\n",
      "processing Prubra_DSM6842.PfamScan\n",
      "loading Prubra_DSM6842.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 1700\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 675\n",
      "writing file Prubra_DSM6842_and_uniprot.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Prubra_DSM6842_and_uniprot.undirected_network.json...\n",
      "completed writing file\n",
      "processing Scoelicolor_A32.PfamScan\n",
      "loading Scoelicolor_A32.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 2938\n",
      "now processing protein #2000 out of 2938\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 775\n",
      "writing file Scoelicolor_A32_and_uniprot.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Scoelicolor_A32_and_uniprot.undirected_network.json...\n",
      "completed writing file\n",
      "processing Pfluorescens_Pf5.PfamScan\n",
      "loading Pfluorescens_Pf5.PfamScan\n",
      "filtering for domains we care about\n",
      "filtering for best hit per clan \n",
      "now processing protein #1000 out of 2264\n",
      "now processing protein #2000 out of 2264\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "drawing edges from node # 500 out of 739\n",
      "writing file Pfluorescens_Pf5_and_uniprot.directed_network.json...\n",
      "completed writing file\n",
      "drawing nodes...\n",
      "drawing edges...\n",
      "writing file Pfluorescens_Pf5_and_uniprot.undirected_network.json...\n",
      "completed writing file\n",
      "Completed.\n"
     ]
    }
   ],
   "source": [
    "file_prefixes=[\"Ecoli_K12_MG1655\",\n",
    "               \"Stropica_CNB-440\",\n",
    "               \"Prubra_DSM6842\",\n",
    "               \"Scoelicolor_A32\",\n",
    "               \"Pfluorescens_Pf5\",\n",
    "              ]\n",
    "\n",
    "input_file_suffix=\".PfamScan\"\n",
    "output_file_suffix=\".json\"\n",
    "\n",
    "\n",
    "for prefix in file_prefixes:\n",
    "    pfamscan_filename=prefix+input_file_suffix\n",
    "    print(\"processing \"+pfamscan_filename)\n",
    "    output_network_filename=prefix+output_file_suffix\n",
    "    domainoccurrences=DomainOccurrences(prefix,*load_PfamScan_domain_pairs_we_care_about(pfamscan_filename))\n",
    "    domainoccurrences.generate_coocurrence_network(prefix+\"_and_uniprot.directed_network.json\",\n",
    "                                                   all_of_uniprot_domainoccurrences, #this argument makes the new magic possible\n",
    "                                                   generate_directed_graph=True)\n",
    "    domainoccurrences.generate_coocurrence_network(prefix+\"_and_uniprot.undirected_network.json\",\n",
    "                                                   all_of_uniprot_domainoccurrences, #this argument makes the new magic possible\n",
    "                                                   generate_directed_graph=False)\n",
    "print(\"Completed.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Using the information about enzyme domains pairs for all of uniprot, we can calculate some general statistics about domain pairs in Uniprot"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1745 total domains analyzed; all possible combinations would make 1521640 pairs\n",
      "For the 30658 domain pairs with at least 1 proteins in Uniprot, 29997 (97.84%) are unannotated (i.e., do not have any entries in MIBiG, BRENDA, or curated as catalytic in Uniprot)\n",
      "For the 3381 domain pairs with at least 10 proteins in Uniprot, 2798 (82.76%) are unannotated (i.e., do not have any entries in MIBiG, BRENDA, or curated as catalytic in Uniprot)\n",
      "For the 902 domain pairs with at least 100 proteins in Uniprot, 504 (55.88%) are unannotated (i.e., do not have any entries in MIBiG, BRENDA, or curated as catalytic in Uniprot)\n",
      "For the 252 domain pairs with at least 1000 proteins in Uniprot, 54 (21.43%) are unannotated (i.e., do not have any entries in MIBiG, BRENDA, or curated as catalytic in Uniprot)\n",
      "For the 24 domain pairs with at least 10000 proteins in Uniprot, 0 (0.00%) are unannotated (i.e., do not have any entries in MIBiG, BRENDA, or curated as catalytic in Uniprot)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "\n",
    "print(\"{} total domains analyzed; all possible combinations would make {} pairs\".format(num_domains,total_combinations))\n",
    "\n",
    "domain_pairs_to_uniprot_frozensets_dict=all_of_uniprot_domainoccurrences.domain_pair_frozenset_to_protein_list_dict\n",
    "number_of_pairs_dict={k:len(v) for k,v in domain_pairs_to_uniprot_frozensets_dict.items()}\n",
    "\n",
    "pairs_in_mibig_or_uniprotcatalytic_or_brenda={k:(uniprot_curated_catalytic_domainoccurrences.get_domain_pair_protein_tally(k)>0) or\n",
    "                                              (mibig_domainoccurrences.get_domain_pair_protein_tally(k)>0) or\n",
    "                                              (brenda_domainoccurrences.get_domain_pair_protein_tally(k)>0) for k in\n",
    "                                              domain_pairs_to_uniprot_frozensets_dict.keys()}\n",
    "\n",
    "number_of_pairs_df=pandas.DataFrame({\"number_of_pairs\":number_of_pairs_dict,\"domain_pair_in_mibig_or_uniprotcatalytic_or_brenda\":pairs_in_mibig_or_uniprotcatalytic_or_brenda})\n",
    "number_of_unannotated_pairs_df=number_of_pairs_df[~number_of_pairs_df[\"domain_pair_in_mibig_or_uniprotcatalytic_or_brenda\"]]\n",
    "\n",
    "order_of_magnitude_index=pandas.IntervalIndex.from_breaks([1,10,100,1000,10000,100000],closed='left')\n",
    "number_of_pairs_binned_by_order_of_magnitude=pandas.cut(number_of_pairs_df[\"number_of_pairs\"],order_of_magnitude_index)\n",
    "number_of_annotated_pairs_binned_by_order_of_magnitude=pandas.cut(number_of_pairs_df[number_of_pairs_df[\"domain_pair_in_mibig_or_uniprotcatalytic_or_brenda\"]][\"number_of_pairs\"],order_of_magnitude_index)\n",
    "number_of_unannotated_pairs_binned_by_order_of_magnitude=pandas.cut(number_of_unannotated_pairs_df[\"number_of_pairs\"],order_of_magnitude_index)\n",
    "\n",
    "plotting_df=pandas.DataFrame({\"all domain pairs\":number_of_pairs_binned_by_order_of_magnitude.value_counts(),\"at least one protein in MIBiG, BRENDA, or curated as catalytic in Uniprot\":number_of_annotated_pairs_binned_by_order_of_magnitude.value_counts(),\"no proteins in MIBiG, BRENDA, or curated as catalytic in Uniprot\":number_of_unannotated_pairs_binned_by_order_of_magnitude.value_counts()})\n",
    "\n",
    "log_ax=plotting_df.plot(kind='bar',logy=True)\n",
    "log_ax.set_ylabel(\"Number of domain pairs (log scale)\")\n",
    "\n",
    "linear_ax=plotting_df.plot(kind='bar',ylim=[0,3000])\n",
    "linear_ax.set_ylabel(\"Number of domain pairs (linear scale)\")\n",
    "\n",
    "for ax in (log_ax,linear_ax):\n",
    "    ax.set_xlabel(\"Number of proteins in Uniprot that contain a pair of domains\")\n",
    "    ax.set_xticklabels([\"{} - {}\".format(i.left,i.right-1) for i in order_of_magnitude_index])\n",
    "    ax.legend(loc='lower center', bbox_to_anchor=(0.5, 1.02))\n",
    "\n",
    "for number in [1,10,100,1000,10000]:\n",
    "    total=len(number_of_pairs_df[number_of_pairs_df[\"number_of_pairs\"]>=number])\n",
    "    unannotated=len(number_of_unannotated_pairs_df[number_of_unannotated_pairs_df[\"number_of_pairs\"]>=number])\n",
    "    print(\"For the {} domain pairs with at least {} proteins in Uniprot, {} ({:.2f}%) are unannotated (i.e., do not have any entries in MIBiG, BRENDA, or curated as catalytic in Uniprot)\".format(total,number,unannotated,100.0*unannotated/total))\n",
    "    \n",
    "    \n",
    "#these results do not change substantially when I include the \"u\" category in the analysis"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}