def benchmark_recalibrators(job, sample, ref, dbsnp, mills):
_log.info("Downloading ref")
ref_id = download_url_job(job, ref)
_log.info("Indexing reference.")
faidx = run_samtools_faidx(job, ref_id)
_log.info("Extracting reference sequence dictionary")
ref_dict = run_picard_create_sequence_dictionary(job, ref_id)
_log.info("Downloading dbSNP VCF")
dbsnp_id = download_url_job(job, dbsnp)
_log.info("Downloading Mills VCF")
mills_id = download_url_job(job, mills)
_log.info("Downloading reads")
reads_id = download_url_job(job, sample)
_log.info("Sorting reads by coordinate.")
coordinate_sorted_bam = run_sambamba_sort(job, reads_id)
_log.info("Indexing sorted BAM.")
bam_index = run_samtools_index(job, coordinate_sorted_bam)
run_adam_bqsr(job, reads_id, dbsnp_id)
run_gatk3_bqsr(job, reads_id, bam_index, ref_id, faidx, ref_dict, dbsnp_id,
mills_id)
def benchmark_realigners(job, sample, ref, g1k, mills):
_log.info("Downloading ref")
ref_id = download_url_job(job, ref)
_log.info("Indexing reference.")
faidx = run_samtools_faidx(job, ref_id)
_log.info("Extracting reference sequence dictionary")
ref_dict = run_picard_create_sequence_dictionary(job, ref_id)
_log.info("Downloading 1000G VCF")
g1k_id = download_url_job(job, g1k)
_log.info("Downloading Mills VCF")
mills_id = download_url_job(job, mills)
_log.info("Downloading reads")
reads_id = download_url_job(job, sample)
_log.info("Sorting reads by coordinate.")
coordinate_sorted_bam = run_sambamba_sort(job, reads_id)
_log.info("Indexing sorted BAM.")
bam_index = run_samtools_index(job, coordinate_sorted_bam)
run_adam_ri(job, reads_id)
run_gatk3_ir(job, reads_id, bam_index, ref_id, faidx, ref_dict, g1k_id,
mills_id)
def benchmark_duplicate_markers(job, sample):
_log.info("Downloading reads")
reads_id = download_url_job(job, sample)
_log.info("Sorting reads by coordinate.")
coordinate_sorted_bam = run_sambamba_sort(job, reads_id)
_log.info("Indexing sorted BAM.")
bam_index = run_samtools_index(job, coordinate_sorted_bam)
_log.info("Marking duplicates with picard.")
picard_bam = picard_mark_duplicates(job, coordinate_sorted_bam, bam_index)
_log.info("Marking duplicates with samtools.")
samtools_bam = run_samtools_rmdup(job, coordinate_sorted_bam)
_log.info("Marking duplicates with sambamba.")
sambamba_bam = run_sambamba_markdup(job, coordinate_sorted_bam)
run_adam_markdups(job, reads_id)
_log.info("Sorting reads by name.")
queryname_sorted_bam = run_sambamba_sort(job, reads_id, sort_by_name=True)
_log.info("Dumping queryname sorted sam to bam.")
queryname_sorted_sam = run_samtools_view(job, queryname_sorted_bam)
_log.info("Marking duplicates with SAMBLASTER.")
samblaster_sam = run_samblaster(job, queryname_sorted_sam)
def gatk3_transform(job, ref, in_file, snp_file, g1k_indels, mills_indels):
_log.info("Downloading ref")
ref_id = download_url_job(job, ref)
_log.info("Indexing reference.")
faidx = run_samtools_faidx(job, ref_id)
_log.info("Extracting reference sequence dictionary")
ref_dict = run_picard_create_sequence_dictionary(job, ref_id)
_log.info("Downloading reads")
reads_id = download_url_job(job, in_file)
_log.info("Sorting reads.")
sorted_bam = run_samtools_sort(job, reads_id)
_log.info("Indexing reads.")
bai = run_samtools_index(job, sorted_bam)
_log.info("Downloading resources")
g1k_id = download_url_job(job, g1k_indels)
mills_id = download_url_job(job, mills_indels)
snp_id = download_url_job(job, snp_file)
_log.info("Running GATK preprocessing")
return run_gatk_preprocessing(job,
sorted_bam,
bai,
ref_id,
ref_dict,
faidx,
g1k_id,
mills_id,
snp_id,
realign=True)
def benchmark_sorters(job, sample):
_log.info("Downloading reads")
reads_id = download_url_job(job, sample)
_log.info("Sorting reads with picard.")
picard_sorted_bam = run_picard_sort(job, reads_id)
_log.info("Sorting reads with samtools.")
samtools_sorted_bam = run_samtools_sort(job, reads_id)
_log.info("Sorting reads with sambamba.")
sambamba_sorted_bam = run_sambamba_sort(job, reads_id)
run_adam_sort(job, reads_id)