Merge pull request #19 from microbiomedata/bbcmsouts

Shortreads bbcms #18

input.json

@@ -1,9 +1,7 @@
 {
-  "jgi_metaASM.input_file":["/global/cfs/projectdirs/m3408/ficus/11809.7.220839.TCCTGAG-ACTGCAT.fastq.gz"],
-  "jgi_metaASM.rename_contig_prefix":"503125_160870",
-  "jgi_metaASM.proj":"nmdc:503125_160870",
-  "jgi_metaASM.resource": "NERSC -- perlmutter",
-  "jgi_metaASM.informed_by": "nmdc:xxxxxx",
-  "jgi_metaASM.memory": "105G",
-  "jgi_metaASM.threads": "16"
+  "jgi_metaAssembly.input_files":["https://data.microbiomedata.org/data/test_data/11809.7.220839.TCCTGAG-ACTGCAT.fastq.gz"],
+  "jgi_metaAssembly.proj":"nmdc:503125_160870",
+  "jgi_metaAssembly.memory": "105G",
+  "jgi_metaAssembly.threads": "16",
+  "jgi_metaAssembly.shortRead": true
 }

jgi_assembly.wdl

@@ -1,7 +1,8 @@
 version 1.0
 import "shortReads_assembly.wdl" as srma
 import "make_interleaved_WDL/make_interleaved_reads.wdl" as int
-import "jgi_meta_wdl/metagenome_improved/metaflye.wdl" as lrma
+# import "jgi_meta_wdl/metagenome_improved/metaflye.wdl" as lrma
+import "https://code.jgi.doe.gov/BFoster/jgi_meta_wdl/-/raw/bc7c4371ea0fa83355bada341ec353b9feb3eff2/metagenome_improved/metaflye.wdl" as lrma
 
 workflow jgi_metaAssembly{
     input {  
@@ -12,7 +13,7 @@ workflow jgi_metaAssembly{
         String? memory
         String? threads
         # longReads parameters
-        Array[File] input_files
+        Array[String] input_files
         String flye_container = "staphb/flye:2.9.2"
         String flye_parameters = "--meta -o flye -t 32 --pacbio-hifi"
         String smrtlink_container = "bryce911/smrtlink:12.0.0.177059"
@@ -99,14 +100,14 @@ workflow jgi_metaAssembly{
         File? sr_covstats=jgi_metaASM.covstats
         File? sr_asmstats=jgi_metaASM.asmstats
         File? sr_asminfo=jgi_metaASM.asminfo
-
+        File? sr_bbcms_fq = jgi_metaASM.bbcms_fastq
           
     }
 }
 
 
 task finish_lrasm {
-    input{
+    input {
     File contigs
     File bam
     File scaffolds

labels.json

@@ -0,0 +1,6 @@
+{
+  "sample_id": "503125_159845",
+  "project_id": "503125",
+  "pipeline": "metaAssembly",
+  "submitter": "vli"
+}

shortReads_assembly.wdl

@@ -4,7 +4,7 @@ workflow jgi_metaASM {
         # String? outdir
         String? memory
         String? threads
-        File? input_file
+        String? input_file
         String proj
         String prefix=sub(proj, ":", "_")
         String rename_contig_prefix="scaffold"
@@ -72,7 +72,8 @@ workflow jgi_metaASM {
         bam=read_mapping_pairs.outbamfile,
         samgz=read_mapping_pairs.outsamfile,
         covstats=read_mapping_pairs.outcovfile,
-        asmstats=create_agp.outstats
+        asmstats=create_agp.outstats,
+        bbcms_fastq = bbcms.out
     }
 
     # call make_output{
@@ -104,6 +105,7 @@ workflow jgi_metaASM {
         File covstats=finish_asm.outcovstats
         File asmstats=finish_asm.outasmstats
         File asminfo=make_info_file.asminfo
+        File bbcms_fastq = finish_asm.outbbcms
     }
 
     meta {
@@ -117,7 +119,7 @@ workflow jgi_metaASM {
 task stage {
    input{ 
    String container
-   File? input_file
+   String? input_file
    String memory = "4G"
    String target = "staged.fastq.gz"
    String output1 = "input.left.fastq.gz"
@@ -193,6 +195,7 @@ task finish_asm {
     File? samgz
     File? covstats
     File asmstats
+    File bbcms_fastq
     String container
     String proj
     String prefix 
@@ -214,6 +217,7 @@ task finish_asm {
         cat ~{scaffold} | sed ~{sed} > ~{prefix}_scaffolds.fna
         cat ~{covstats} | sed ~{sed} > ~{prefix}_covstats.txt
         cat ~{agp} | sed ~{sed} > ~{prefix}_assembly.agp
+        ln ~{bbcms_fastq} ~{prefix}_bbcms.fastq.gz || ln -s ~{bbcms_fastq} ~{prefix}_bbcms.fastq.gz
 
        ## Bam file     
        samtools view -h ~{bam} | sed ~{sed} | \
@@ -234,6 +238,7 @@ task finish_asm {
         File outsamgz = "~{prefix}_pairedMapped.sam.gz"
         File outcovstats = "~{prefix}_covstats.txt"
         File outasmstats = "stats.json"
+        File outbbcms = "~{prefix}_bbcms.fastq.gz"
     }
 
     runtime {

submit_curl.sh

@@ -0,0 +1,15 @@
+#!/bin/bash
+#
+
+
+# zip bundle.zip *wdl
+WD=/global/cfs/cdirs/m3408/aim2/dev/kli_training/mg_annotation/
+
+curl --netrc -X POST "https://nmdc-cromwell.freeddns.org:8443/api/workflows/v1" \
+	-H "accept: application/json" \
+	-H "Content-Type: multipart/form-data" \
+	-F "workflowSource=@jgi_assembly.wdl" \
+	-F "workflowInputs=@input.json;type=application/json"  \
+    -F "workflowDependencies=@imports.zip" \
+	-F "labels=@labels.json;type=application/json"
+