expose intrahost optional params in config.yaml

This addresses #412 and exposes the following parameters in the config
file:
intrahost.py vphaser_one_sample --minReadsEach and --maxBias
intrahost.py merge_to_vcf --naive_filter

intrahost.py

@@ -328,7 +328,7 @@ def parser_vphaser_one_sample(parser=argparse.ArgumentParser()):
     parser.add_argument("--removeDoublyMappedReads",
                         default=False,
                         action="store_true",
-                        help="""When calling V-Phaser, keep reads mapping to more than one contig.""")
+                        help="""When calling V-Phaser, remove reads mapping to more than one contig. Default is to keep the reads.""")
     util.cmd.common_args(parser, (('loglevel', None), ('version', None)))
     util.cmd.attach_main(parser, vphaser_one_sample, split_args=True)
     return parser

pipes/config.yaml

@@ -154,6 +154,19 @@ mafft_maxiters: 1000
 # See: https://github.com/trinityrnaseq/trinityrnaseq/wiki/Trinity-FAQ#ques_why_so_many_transcripts
 trinity_n_reads: 250000
 
+# Minimum number of reads on each strand
+vphaser_min_reads_each: 5
+
+# Maximum allowable ratio of number of reads on the two
+# strands. Ignored if vphaser_max_bins=0.
+vphaser_max_bins: 10
+
+# A simple filter for the VCF merge step.
+# If set to true, keep only the alleles that have at least two
+# independent libraries of support and 
+# allele freq > 0.005. If false, no filtering is performed.
+vcf_merge_naive_filter: false
+
 #    |----------------------- Data storage locations ---------------------------
 
 # The parent directory containing data sub-directories.

pipes/rules/intrahost.rules

@@ -23,10 +23,12 @@ rule isnvs_per_sample:
     params: LSF=config.get('LSF_queues', {}).get('short', '-W 4:00'),
             UGER=config.get('UGER_queues', {}).get('short', '-q short'),
             logid="{sample}",
-            numThreads=str(config.get("number_of_threads", 1))
+            numThreads=str(config.get("number_of_threads", 1)),
+            minReadsPerStrand="--minReadsEach {}".format(config['vphaser_min_reads_each']) if 'vphaser_min_reads_each' in config else "",
+            maxBias="--maxBias {}".format(config['vphaser_max_bins']) if 'vphaser_max_bins' in config else ""
     run:
             makedirs(config["data_dir"]+'/'+config["subdirs"]["intrahost"])
-            shell("{config[bin_dir]}/intrahost.py vphaser_one_sample {input} {output} --vphaserNumThreads {params.numThreads} --removeDoublyMappedReads")
+            shell("{config[bin_dir]}/intrahost.py vphaser_one_sample {input} {output} --vphaserNumThreads {params.numThreads} --removeDoublyMappedReads {params.minReadsPerStrand} {params.maxBias}")
 
 rule isnvs_vcf:
     input:
@@ -87,14 +89,16 @@ rule isnvs_vcf_filtered:
             refGenome=os.path.join(config["ref_genome_dir"],"reference"+".fasta"),
             snpEff_ref=" ".join(config["accessions_for_ref_genome_build"]),
             samples=list(read_samples_file(config["samples_assembly"])),
-            emailAddress=config["email_point_of_contact_for_ncbi"]
+            emailAddress=config["email_point_of_contact_for_ncbi"],
+            naiveFilter="--naive_filter" if config["email_point_of_contact_for_ncbi"] else ""
     run:
             shell("{config[bin_dir]}/intrahost.py merge_to_vcf {params.refGenome} {output[0]}"
                 + " --samples " + " ".join(params.samples)
                 + " --isnvs " + " ".join(["{config[data_dir]}/{config[subdirs][intrahost]}/vphaser2."+s+".txt.gz" for s in params.samples])
                 + " --alignments " + " ".join(["{config[data_dir]}/{config[subdirs][multialign_ref]}/aligned_" + str(n) + ".fasta" for n in range(1, len(config["accessions_for_ref_genome_build"])+1)])
                 + " --strip_chr_version"
-                + "--parse_accession" # the vcf chr column must match a chr known to snpEff; we have an option to parse out only the accession
+                + " {params.naiveFilter}"
+                + " --parse_accession" # the vcf chr column must match a chr known to snpEff; we have an option to parse out only the accession
                 )
             shell("{config[bin_dir]}/interhost.py snpEff {output[0]} {params.snpEff_ref} {output[1]} {params.emailAddress}")
             shell("{config[bin_dir]}/intrahost.py iSNV_table {output[1]} {output[2]}")