fix count accessions by taxid test

chanzuckerberg · Apr 8, 2024 · 2e1c117 · 2e1c117
1 parent 890e28d
commit 2e1c117
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 1 deletion.
diff --git a/workflows/index-generation/index-generation.wdl b/workflows/index-generation/index-generation.wdl
@@ -495,7 +495,7 @@ task CompressDatabase {
             --input-taxid-dir $READS_BY_TAXID_PATH \
             --output-taxid-dir $SORTED_TAXID_DIR_NAME
 
-        mkdir $SPLIT_APART_TAXID_DIR_NAME
+        mkdir $SPLIT_APART_TAXID_DIR_NAME # this is needed to split up large taxids into smaller chunks to reduce memory usage
 
         ncbi-compress fasta-compress-from-taxid-dir  ~{if database_type == "nr" then "--is-protein-fasta" else ""} \
             --input-fasta-dir $SORTED_TAXID_DIR_NAME \

diff --git a/workflows/index-generation/ncbi-compress/src/fasta_tools.rs b/workflows/index-generation/ncbi-compress/src/fasta_tools.rs
@@ -70,6 +70,7 @@ pub mod fasta_tools {
                 .open(&output_tsv_path)?;
 
             writer.write_all(format!("{}\t{}\n", taxid, records.count()).as_bytes())?;
+            writer.flush()?;
         }
         Ok(())
     }
@@ -288,6 +289,7 @@ mod tests {
         let mut truth_rdr = ReaderBuilder::new()
             .delimiter(b'\t')
             .trim(Trim::All)
+            .has_headers(false)
             .from_path(output_truth_tsv_file)
             .unwrap();
         let mut truth_records = Vec::new();
@@ -305,6 +307,7 @@ mod tests {
         let mut test_rdr = ReaderBuilder::new()
             .delimiter(b'\t')
             .trim(Trim::All)
+            .has_headers(false)
             .from_path(test_truth_tsv_file_path_str)
             .unwrap();
         let mut test_records = Vec::new();