revert BARCODE_PARAMS stuff and make sure to pass read_structure to i…

…llumina_basecalls in snakemake
broadinstitute · Apr 29, 2015 · 00e34df · 00e34df
1 parent e6041dd
commit 00e34df
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 15 deletions.
diff --git a/broad_utils.py b/broad_utils.py
@@ -188,14 +188,13 @@ def make_params_file(inFile, bamDir, outFile):
     if any(row.get('barcode_2') for row in util.file.read_tabfile_dict(inFile)):
         header = ['OUTPUT', 'SAMPLE_ALIAS', 'LIBRARY_NAME', 'BARCODE_1', 'BARCODE_2']
     else:
-        header = ['BARCODE', 'OUTPUT', 'SAMPLE_ALIAS', 'LIBRARY_NAME']
+        header = ['OUTPUT', 'SAMPLE_ALIAS', 'LIBRARY_NAME', 'BARCODE_1']
     with open(outFile, 'wt') as outf:
         outf.write('\t'.join(header)+'\n')
         rows = list(util.file.read_tabfile_dict(inFile))
         rows.append({'barcode_1':'N','barcode_2':'N','sample':'Unmatched'})
         for row in rows:
-            out  = {'BARCODE':row['barcode_1'],
-                    'BARCODE_1':row['barcode_1'],
+            out  = {'BARCODE_1':row['barcode_1'],
                     'BARCODE_2':row.get('barcode_2',''),
                     'SAMPLE_ALIAS':row['sample'],
                     'LIBRARY_NAME':row['sample']}

diff --git a/pipes/rules/demux.rules b/pipes/rules/demux.rules
@@ -126,7 +126,11 @@ rule illumina_basecalls:
             lane = get_one_lane_from_run(wildcards.flowcell, wildcards.lane, config['seqruns_demux'])
             dir = lane['bustard_dir']
             run_date = lane.get('seq_run_date')
-            shell("{config[binDir]}/broad_utils.py illumina_basecalls {dir} {input[1]} {wildcards.flowcell} {wildcards.lane} {input[0]} --include_non_pf_reads=false --run_start_date={run_date} --sequencing_center={params.center}")
+            opts = ''
+            for opt in ('read_structure',):
+                if lane.get(opt):
+                    opts += ' --%s=%s' % (opt, lane[opt])
+            shell("{config[binDir]}/broad_utils.py illumina_basecalls {dir} {input[1]} {wildcards.flowcell} {wildcards.lane} {input[0]} --include_non_pf_reads=false --run_start_date={run_date} --sequencing_center={params.center} {opts}")
 
 def demux_move_bams_inputs(wildcards):
     lane = get_one_lane_from_run(wildcards.flowcell, wildcards.lane, config.get('seqruns_demux',''))
@@ -143,15 +147,15 @@ rule move_bams_demux:
             shutil.move(input[0], output[0])
 
 rule bams_from_fastq:
-    input:  os.path.join(config['dataDir'],config['subdirs']['source'],'{sample}_R1_{idx}.fastq'),
-            os.path.join(config['dataDir'],config['subdirs']['source'],'{sample}_R2_{idx}.fastq')
-    output: os.path.join(config['dataDir'],config['subdirs']['source'],'{sample}_{idx}.bam')
+    input:  os.path.join(config['dataDir'],config['subdirs']['source'],'{sample}_L001_R1_001.fastq.gz'),
+            os.path.join(config['dataDir'],config['subdirs']['source'],'{sample}_L001_R2_001.fastq.gz')
+    output: os.path.join(config['dataDir'],config['subdirs']['source'],'{sample}.bam')
     params: LSF=config.get('LSF_queues', {}).get('short', '-W 4:00'),
-            logid="{sample}_{idx}",
+            logid="{sample}",
             center=config["seq_center"]
     run:
             makedirs(os.path.join(config['dataDir'], config['subdirs']['source']))
-            shell("{config[binDir]}/read_utils.py fastq_to_bam {input} {output} --sampleName {wildcards.sample} --picardOptions PLATFORM=illumina SEQUENCING_CENTER={params.center} LIBRARY_NAME={wildcards.sample}_{wildcards.idx} SORT_ORDER=queryname")
+            shell("{config[binDir]}/read_utils.py fastq_to_bam {input} {output} --sampleName {wildcards.sample} --picardOptions PLATFORM=illumina SEQUENCING_CENTER={params.center} LIBRARY_NAME={wildcards.sample} SORT_ORDER=queryname")
 
 ruleorder: move_bams_demux > bams_from_fastq
 
diff --git a/tools/picard.py b/tools/picard.py
@@ -212,11 +212,7 @@ def execute(self, basecalls_dir, barcodes_dir,
         opts += ['BASECALLS_DIR='+basecalls_dir,
                 'BARCODES_DIR='+barcodes_dir,
                 'LANE='+str(lane),
-                'RUN_BARCODE='+run_barcode]
-        if 'BARCODE' in util.file.readFlatFileHeader(library_params):
-            # yes, it's deprecated, but it's the only way to do single-barcode pools
-            opts.append('BARCODE_PARAMS='+library_params)
-        else:
-            opts.append('LIBRARY_PARAMS='+library_params)
+                'RUN_BARCODE='+run_barcode,
+                'LIBRARY_PARAMS='+library_params]
         PicardTools.execute(self, self.subtoolName, opts, JVMmemory)