Merge pull request #38 from aofarrel/california-rework

California rework (CDC decontamination reference)

inputs/myco_raw_local.json

@@ -1,6 +1,6 @@
 {
   "myco.paired_fastq_sets": [["./inputs/data/fastqs/ERR7057565_tiny_1.fq", "./inputs/data/fastqs/ERR7057565_tiny_2.fq"]],
-  "myco.decorate_tree":true,
+  "myco.tree_decoration":true,
   "myco.timeout_decontam_part1":0,
   "myco.timeout_decontam_part2":0,
   "myco.timeout_variant_caller":0

inputs/myco_raw_terra_datatable.json

@@ -0,0 +1,7 @@
+{
+  "myco.paired_fastq_sets": [[this.fq_1, this.fq_2]],
+  "myco.decorate_tree":true,
+  "myco.timeout_decontam_part1":0,
+  "myco.timeout_decontam_part2":0,
+  "myco.timeout_variant_caller":0
+}

inputs/terra_datatable.tsv

@@ -0,0 +1,18 @@
+entity:Multistrain_BioSample_id	TBProfiler_lineage	dataset	fq_1	fq_2
+SAMEA104362188	lineage4.3.3;lineage2.2.1	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA104362188_ERR2184347_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA104362188_ERR2184347_2.fastq
+SAMEA104362488	lineage4.1.1;lineage2.2.1	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA104362488_ERR2184647_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA104362488_ERR2184647_2.fastq
+SAMEA104600448	lineage4.8;lineage2.2.1	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA104600448_ERR2307669_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA104600448_ERR2307669_2.fastq
+SAMEA110024222	lineage4.3.2;lineage4.1.2.1	partytime	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA110024222_ERR9786926_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA110024222_ERR9786926_2.fastq
+SAMEA110037041	lineage4.7;lineage4.1.2.1;lineage2.2.1	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA110037041_ERR9786506_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA110037041_ERR9786506_2.fastq
+SAMEA110046394	lineage4.3.2;lineage4.1.2.1;lineage2.2.1	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA110046394_ERR9786640_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA110046394_ERR9786640_2.fastq
+SAMEA1101881	lineage4.8;lineage3	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA1101881_ERR2510523_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA1101881_ERR2510523_2.fastq
+SAMEA3257650	lineage4.3.2.1;lineage2.2.1	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA3257650_ERR2515823_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA3257650_ERR2515823_2.fastq
+SAMEA4441507	lineage4.3.2.1;lineage2.2.1.1	rand03333	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA4441507_ERR1633857_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA4441507_ERR1633857_2.fastq
+SAMEA7546014	lineage4.3.3;lineage4.1.1	rand13000_part1	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA7546014_ERR4812510_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA7546014_ERR4812510_2.fastq
+SAMEA7562370	lineage4.1.1.2;lineage3	rand13000_part2	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA7562370_ERR4829931_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMEA7562370_ERR4829931_2.fastq
+SAMN15772665	lineage4.3.4.1;lineage2.2	rand09999	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN15772665_SRR12420677_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN15772665_SRR12420677_2.fastq
+SAMN25995062	lineage4.3.4;lineage3.1.1;lineage1.1.2	rand04000	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN25995062_SRR19428603_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN25995062_SRR19428603_2.fastq
+SAMN25995075	lineage4.3.4.2.1;lineage3.1.1;lineage1	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN25995075_SRR19428557_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN25995075_SRR19428557_2.fastq
+SAMN25995080	lineage4.2.2;lineage3.1.1;lineage1.1	rand08888	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN25995080_SRR19428551_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN25995080_SRR19428551_2.fastq
+SAMN30547680	lineage4;lineage2.2.2	rand08888	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN30547680_SRR21279381_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN30547680_SRR21279381_2.fastq
+SAMN30547732	lineage2.2.2;lineage2.2.1	rand12344	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN30547732_SRR21279462_1.fastq	gs://ucsc-pathogen-genomics-public/tb/fq/SAMN30547732_SRR21279462_2.fastq

myco_raw.wdl

@@ -1,7 +1,7 @@
 version 1.0
 
-import "https://raw.githubusercontent.com/aofarrel/clockwork-wdl/2.10.0/tasks/combined_decontamination.wdl" as clckwrk_combonation
-import "https://raw.githubusercontent.com/aofarrel/clockwork-wdl/2.10.0/tasks/variant_call_one_sample.wdl" as clckwrk_var_call
+import "https://raw.githubusercontent.com/aofarrel/clockwork-wdl/2.11.0/tasks/combined_decontamination.wdl" as clckwrk_combonation
+import "https://raw.githubusercontent.com/aofarrel/clockwork-wdl/2.11.0/tasks/variant_call_one_sample.wdl" as clckwrk_var_call
 import "https://raw.githubusercontent.com/aofarrel/SRANWRP/v1.1.12/tasks/processing_tasks.wdl" as sranwrp_processing
 import "https://raw.githubusercontent.com/aofarrel/tree_nine/0.0.10/tree_nine.wdl" as build_treesWF
 import "https://raw.githubusercontent.com/aofarrel/parsevcf/1.2.0/vcf_to_diff.wdl" as diff
@@ -17,6 +17,7 @@ workflow myco {
 		Int     covstatsQC_minimum_coverage    =   10
 		Int     covstatsQC_max_percent_unmapped=    2
 		Boolean covstatsQC_skip_entirely       = false
+		Boolean decontam_use_CDC_varpipe_ref   = true
 		File?   diffQC_mask_bedfile
 		Int     diffQC_max_percent_low_coverage=    20
 		Int     diffQC_low_coverage_cutoff     =    10
@@ -79,6 +80,7 @@ workflow myco {
 	scatter(paired_fastqs in paired_fastq_sets) {
 		call clckwrk_combonation.combined_decontamination_single_ref_included as decontam_each_sample {
 			input:
+				docker_image = if decontam_use_CDC_varpipe_ref then "ashedpotatoes/clockwork-plus:v0.11.3.7-CDC" else "ashedpotatoes/clockwork-plus:v0.11.3.2-full",
 				unsorted_sam = true,
 				reads_files = paired_fastqs,
 				subsample_cutoff = subsample_cutoff,
@@ -318,7 +320,7 @@ workflow myco {
 
 			# if there is only one sample, there's no need to run tasks
 			if(length(paired_fastq_sets) == 1) {
-				Int    single_sample_tbprof_bam_depth      = read_int(coerced_bam_depths[0])
+				Int    single_sample_tbprof_bam_depth      = coerced_bam_depths[0]
 				String single_sample_tbprof_bam_resistance = coerced_bam_resistances[0]
 				String single_sample_tbprof_bam_strain     = coerced_bam_strains[0]
 			}