From 76883d02e6073f990ea980dad1f8cf21121255ff Mon Sep 17 00:00:00 2001
From: Peter Amstutz
Date: Mon, 22 Jun 2020 18:23:40 +0000
Subject: Adjust QC filter and relabel output sequence with sample_id

---
 bh20seqanalyzer/main.py                      | 9 ++++++---
 workflows/fastq2fasta/bam2fasta.cwl          | 2 ++
 workflows/fastq2fasta/bcftools-consensus.cwl | 9 ++++++++-
 workflows/fastq2fasta/bcftools-view-qc.cwl   | 2 +-
 workflows/fastq2fasta/fastq2fasta.cwl        | 2 ++
 5 files changed, 19 insertions(+), 5 deletions(-)

diff --git a/bh20seqanalyzer/main.py b/bh20seqanalyzer/main.py
index 9164190..9a36cae 100644
--- a/bh20seqanalyzer/main.py
+++ b/bh20seqanalyzer/main.py
@@ -30,6 +30,7 @@ def validate_upload(api, collection, validated_project,
         try:
             metadata_content = ruamel.yaml.round_trip_load(col.open("metadata.yaml"))
             metadata_content["id"] = "http://arvados.org/keep:%s/metadata.yaml" % collection["portable_data_hash"]
+            sample_id = metadata_content["sample"]["sample_id"]
             add_lc_filename(metadata_content, metadata_content["id"])
             valid = qc_metadata(metadata_content) and valid
         except Exception as e:
@@ -51,7 +52,7 @@ def validate_upload(api, collection, validated_project,
                         logging.info("Expected %s but magic says it should be %s", n, tgt)
                         valid = False
                     elif tgt in ("reads.fastq", "reads.fastq.gz", "reads_1.fastq", "reads_1.fastq.gz"):
-                        start_fastq_to_fasta(api, collection, fastq_project, fastq_workflow_uuid, n)
+                        start_fastq_to_fasta(api, collection, fastq_project, fastq_workflow_uuid, n, sample_id)
                         return False
             if tgt is None:
                 valid = False
@@ -108,7 +109,8 @@ def run_workflow(api, parent_project, workflow_uuid, name, inputobj):
 def start_fastq_to_fasta(api, collection,
                          analysis_project,
                          fastq_workflow_uuid,
-                         tgt):
+                         tgt,
+                         sample_id):
 
     params = {
         "metadata": {
@@ -118,7 +120,8 @@ def start_fastq_to_fasta(api, collection,
         "ref_fasta": {
             "class": "File",
             "location": "keep:ffef6a3b77e5e04f8f62a7b6f67264d1+556/SARS-CoV2-NC_045512.2.fasta"
-        }
+        },
+        "sample_id": sample_id
     }
 
     if tgt.startswith("reads.fastq"):
diff --git a/workflows/fastq2fasta/bam2fasta.cwl b/workflows/fastq2fasta/bam2fasta.cwl
index efe580f..dd4020b 100644
--- a/workflows/fastq2fasta/bam2fasta.cwl
+++ b/workflows/fastq2fasta/bam2fasta.cwl
@@ -15,6 +15,7 @@ inputs:
   threads:
     type: int
     default: 4
+  sample_id: string
 
 outputs:
   out_fasta:
@@ -61,5 +62,6 @@ steps:
     in:
       ref_fasta: fasta
       vcf: bcftools_index_after_qc/indexed
+      sample_id: sample_id
     out: [out_fasta]
     run: bcftools-consensus.cwl
diff --git a/workflows/fastq2fasta/bcftools-consensus.cwl b/workflows/fastq2fasta/bcftools-consensus.cwl
index c111792..dffdbe3 100644
--- a/workflows/fastq2fasta/bcftools-consensus.cwl
+++ b/workflows/fastq2fasta/bcftools-consensus.cwl
@@ -4,20 +4,27 @@ cwlVersion: v1.1
 hints:
   DockerRequirement:
     dockerPull: "quay.io/biocontainers/bcftools:1.10.2--hd2cd319_0"
+  ShellCommandRequirement: {}
 baseCommand: bcftools
 arguments:
   - consensus
-  - -i'QUAL > 1 && GT="A"'
+  - -i
+  - 'QUAL > 1 && GT="a"'
   - -Hla
   - -f
   - $(inputs.ref_fasta)
   - $(inputs.vcf)
+  - {shellQuote: false, valueFrom: "|"}
+  - sed
+  - "s/^>.*/>$(inputs.sample_id)/g"
 inputs:
   - id: ref_fasta
     type: File
   - id: vcf
     type: File
     secondaryFiles: [.csi]
+  - id: sample_id
+    type: string
 outputs:
   - id: out_fasta
     type: stdout
diff --git a/workflows/fastq2fasta/bcftools-view-qc.cwl b/workflows/fastq2fasta/bcftools-view-qc.cwl
index 477c596..336f455 100644
--- a/workflows/fastq2fasta/bcftools-view-qc.cwl
+++ b/workflows/fastq2fasta/bcftools-view-qc.cwl
@@ -8,7 +8,7 @@ baseCommand: bcftools
 arguments:
   - view
   - -i
-  - 'QUAL>1 && (GT="AA" || GT="Aa")'
+  - 'QUAL > 1 && GT="a"'
   - -Oz
   - --threads=$(inputs.threads)
   - $(inputs.bcf)
diff --git a/workflows/fastq2fasta/fastq2fasta.cwl b/workflows/fastq2fasta/fastq2fasta.cwl
index 0cf5c48..d529d99 100644
--- a/workflows/fastq2fasta/fastq2fasta.cwl
+++ b/workflows/fastq2fasta/fastq2fasta.cwl
@@ -22,6 +22,7 @@ inputs:
     type: int
     default: 4
   metadata: File?
+  sample_id: string
 
 outputs:
   out_fasta:
@@ -57,5 +58,6 @@ steps:
       bam: samtools-sort/sorted_bam
       fasta: ref_fasta
       threads: threads
+      sample_id: sample_id
     out: [out_fasta]
     run: bam2fasta.cwl
-- 
cgit v1.2.3