subsample with seqtk instead of using head (#371)

chanzuckerberg · Jun 20, 2024 · a7c54be · a7c54be
1 parent 15f611d
commit a7c54be
Showing 1 changed file with 3 additions and 1 deletion.
diff --git a/workflows/long-read-mngs/run.wdl b/workflows/long-read-mngs/run.wdl
@@ -268,7 +268,9 @@ task RunSubsampling {
 
     command <<<
         set -euxo pipefail
-        head -"~{subsample_depth}" "~{input_fastq}" > sample.subsampled.fastq
+
+        # set seed to 42 for reproducibility
+        seqtk sample -s42 "~{input_fastq}" "~{subsample_depth}" > sample.subsampled.fastq
 
         # We should always have reads after subsampling, but adding for consistency with other steps
         filter_count sample.subsampled.fastq subsampled "No reads remaining after subsampling"