move initial collate step upstream (allows subssampling to be retaine…

…d in realignment)
wtsi-npg · Mar 4, 2020 · c1a1865 · c1a1865
1 parent cfdaa9c
commit c1a1865
Show file tree

Hide file tree

Showing 2 changed files with 49 additions and 45 deletions.
diff --git a/data/vtlib/alignment_wtsi_stage2_template.json b/data/vtlib/alignment_wtsi_stage2_template.json
@@ -119,7 +119,39 @@
 	{
 		"id":"spatial_filter_stats",
 		"type":"OUTFILE",
-		"name":{"subst":"spatial_filter_stats", "required":true, "ifnull":{"subst_constructor":{"vals":[ {"subst":"outdatadir"}, "/", {"subst":"rpt"}, ".spatial_filter.stats" ], "postproc":{"op":"concat","pad":""}}}} },
+		"name":{"subst":"spatial_filter_stats", "required":true, "ifnull":{"subst_constructor":{"vals":[ {"subst":"outdatadir"}, "/", {"subst":"rpt"}, ".spatial_filter.stats" ], "postproc":{"op":"concat","pad":""}}}}
+	},
+	{
+		"id":"collate",
+		"type":"EXEC",
+		"use_STDIN": true,
+		"use_STDOUT": true,
+		"comment": "if realignment_switch is set to 1, will use samtools collate otherwise biobambam collate. Unless the collate_or_namesort parameter has been set directly",
+		"cmd":{
+			"select":"collate_or_namesort", "select_range":[1], "default":{"select":"realignment_switch", "select_range":[1], "default":0,"cases":["collate","stcollate"]},
+			"cases":{
+				"collate": ["bamcollate2", "collate=1", "level=0"],
+				"namesort": [
+						{"subst":"samtools_executable", "required":true, "ifnull":"samtools"}, "sort", "-n",
+						"-l", {"subst":"name_sort_compression","required":true,"ifnull":["0"]},
+						{"subst":"name_sort_mpt_flag","ifnull":{"subst_constructor":{"vals":["-m", {"subst":"coord_sort_mem_per_thread","required":false}]}}},
+						"--threads", {"subst":"name_sort_threads","required":true,"ifnull":4},
+						{"subst":"name_sort_extra_flags", "required":false},
+						"-"
+				],
+				"stcollate": [
+						{"subst":"samtools_executable", "required":true, "ifnull":"samtools"}, "collate", "-O",
+						"-l", {"subst":"stcollate_compression","required":true,"ifnull":["0"]},
+						{"subst":"stcollate_tfn_flag","ifnull":{"subst_constructor":{"vals":["-n", {"subst":"stcollate_tmpfilenum","required":false}]}}},
+						"--threads", {"subst":"stcollate_threads","required":true,"ifnull":4},
+						{"subst":"stcollate_extra_flags", "required":false},
+						"-",
+						{"subst":"stcollate_tmp_prefix", "required":false, "ifnull":"./coltmp"}
+				]
+			}
+		},
+		"comment":"ensure BAM records are gathered by template i.e. queryname"
+	},
 	{
 		"id":"subsample_fqc",
 		"type":"VTFILE",
@@ -155,20 +187,22 @@
 						]
 					}
 				]},
-				"--output",
-				{"packflag":[
-					{"port":"phix_bam_out","direction":"out"},
-					{"select":"no_target_alignment", "default":0, "select_range":[1],
-						"cases":[
-							[",", {"port":"target_bam_out","direction":"out"}],
-							[]
-						]
-					}
-				]},
+				[
+					"--output",
+					{"packflag":[
+						{"port":"phix_bam_out","direction":"out"},
+						{"select":"no_target_alignment", "default":0, "select_range":[1],
+							"cases":[
+								[",", {"port":"target_bam_out","direction":"out"}],
+								[]
+							]
+						}
+					]}
+				],
 				{"select":"no_target_alignment", "default":0, "select_range":[1],
 					"cases":[
 						[],
-						[ "-n", {"port":"target_bam_out","direction":"out"} ]
+						[ {"select":"realignment_switch", "default":0, "select_range":[1], "cases":[ "-n", "-o" ]}, {"port":"target_bam_out","direction":"out"} ]
 					]
 				},
 				"-m", {"port":"af_metrics","direction":"out"}
@@ -209,7 +243,8 @@
 "edges":[
 	{ "id":"src_to_bc2", "from":"crammerge", "to":"spatial_filter" },
 	{ "id":"apply_filter_to_sfqc", "from":"spatial_filter:apply_stats", "to":"spatial_filter_stats" },
-	{ "id":"sf_to_bc2", "from":"spatial_filter", "to":"subsample_fqc" },
+	{ "id":"sf_to_collate", "from":"spatial_filter", "to":"collate" },
+	{ "id":"collate_to_ss", "from":"collate", "to":"subsample_fqc" },
 	{ "id":"ssfqc_to_tgtaln", "from":"subsample_fqc:straight_through1", "to":"target_alignment" },
 	{ "id":"target_alignment_to_alignment_filter", "from":"target_alignment:aligned_to_target", "to":"alignment_filter:target_bam_in" },
 	{ "id":"iab_to_alignment_filter", "from":"target_alignment:aligned_to_phix", "to":"alignment_filter:phix_bam_in" },

diff --git a/data/vtlib/target_alignment.json b/data/vtlib/target_alignment.json
@@ -4,7 +4,7 @@
 "subgraph_io":{
 	"ports":{
 		"inputs":{
-			"_stdin_":"bamcollate2"
+			"_stdin_":"tee3"
 		},
 		"outputs":{
 			"aligned_to_target":"bam12auxmerge",
@@ -32,36 +32,6 @@
 	}
 ],
 "nodes":[
-	{
-		"id":"bamcollate2",
-		"type":"EXEC",
-		"use_STDIN": true,
-		"use_STDOUT": true,
-		"comment": "if realignment_switch is set to 1, will use samtools collate otherwise biobambam collate. Unless the collate_or_namesort parameter has been set directly",
-		"cmd":{
-			"select":"collate_or_namesort", "select_range":[1], "default":{"select":"realignment_switch", "select_range":[1], "default":0,"cases":["collate","stcollate"]},
-			"cases":{
-				"collate": ["bamcollate2", "collate=1", "level=0"],
-				"namesort": [
-						{"subst":"samtools_executable", "required":true, "ifnull":"samtools"}, "sort", "-n",
-						"-l", {"subst":"name_sort_compression","required":true,"ifnull":["0"]},
-						{"subst":"name_sort_mpt_flag","ifnull":{"subst_constructor":{"vals":["-m", {"subst":"coord_sort_mem_per_thread","required":false}]}}},
-						"--threads", {"subst":"name_sort_threads","required":true,"ifnull":4},
-						{"subst":"name_sort_extra_flags", "required":false},
-						"-"
-				],
-				"stcollate": [
-						{"subst":"samtools_executable", "required":true, "ifnull":"samtools"}, "collate", "-O",
-						"-l", {"subst":"stcollate_compression","required":true,"ifnull":["0"]},
-						{"subst":"stcollate_tfn_flag","ifnull":{"subst_constructor":{"vals":["-n", {"subst":"stcollate_tmpfilenum","required":false}]}}},
-						"--threads", {"subst":"stcollate_threads","required":true,"ifnull":4},
-						{"subst":"stcollate_extra_flags", "required":false},
-						"-"
-				]
-			}
-		},
-		"comment":"ensure BAM records are gathered by template i.e. queryname"
-	},
 	{
 		"id":"tee3",
 		"type":"VTFILE",
@@ -186,7 +156,6 @@
 	}
 ],
 "edges":[
-	{ "id":"bc2_to_t3", "from":"bamcollate2", "to":"tee3" },
 	{ "id":"phixaln_to_amp", "from":"tee3:amp", "to":"auxmerge_prep" },
 	{ "id":"amp_to_auxmrg", "from":"auxmerge_prep", "to":"bam12auxmerge:no_aln_bam" },
 	{ "id":"tgtaln_to_prealn", "from":"tee3:prealn", "to":"pre_alignment" },