Merge pull request #103 from nf-core/100-refactoring-input-truth-files

100 refactoring input truth files
nf-core · Nov 6, 2024 · 0ca66e3 · 0ca66e3
2 parents 14812e9 + bf091c6
commit 0ca66e3
Show file tree

Hide file tree

Showing 37 changed files with 374 additions and 763 deletions.
diff --git a/assets/datavzrd/sompy.datavzrd.template.yaml b/assets/datavzrd/sompy.datavzrd.template.yaml
@@ -9,10 +9,12 @@ views:
       columns:
         Tool:
           display-mode: normal
-        Threshold:
+        Type:
           display-mode: normal
         TP_base:
           display-mode: normal
+        TP:
+          display-mode: normal
         FN:
           display-mode: normal
         TP_call:
@@ -23,5 +25,17 @@ views:
           display-mode: normal
         Recall:
           display-mode: normal
-        F1:
+        recall_lower:
+          display-mode: normal
+        recall_upper:
+          display-mode: normal
+        recall2:
+          display-mode: normal
+        precision_lower:
+          display-mode: normal
+        precision_upper:
+          display-mode: normal
+        fp.region.size:
+          display-mode: normal
+        fp.rate:
           display-mode: normal
diff --git a/assets/samplesheet.csv b/assets/samplesheet.csv
diff --git a/assets/samplesheet_full.csv b/assets/samplesheet_full.csv
diff --git a/assets/samplesheet_full_small.csv b/assets/samplesheet_full_small.csv
@@ -0,0 +1,3 @@
+id,test_vcf,caller,refdist,chunksize,normshift,normdist,normsizediff,maxdist,typeignore,dup_to_ins,pctsize,pctseq,pctovl,evaluationmode,subsample
+test1,"https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/data/AshkenazimTrio/analysis/SevenBridges_GraphGATKRefine_05052017/HG002-NA24385-50x.union_170414.split.vcf.gz",graph,,,,,,,,,,,,,
+test2,"https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/data/AshkenazimTrio/analysis/OsloUniversityHospital_Exome_GATK_jointVC_11242015/HG002-HG003-HG004.jointVC.filter.vcf",gatk,,,,,,,,,,,,,"Sample_Diag-excap51-HG002-EEogPU"
diff --git a/assets/samplesheet_full_sv.csv b/assets/samplesheet_full_sv.csv
@@ -0,0 +1,3 @@
+id,test_vcf,caller,refdist,chunksize,normshift,normdist,normsizediff,maxdist,typeignore,dup_to_ins,pctsize,pctseq,pctovl,evaluationmode,subsample
+test1,"https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/release/AshkenazimTrio/HG002_NA24385_son/NIST_SV_v0.6/GIAB_Evaluations/BoutrosLab-SV_curation_GIABv0.6/manta_GIABv0.6/diploidSV_PASS_DUPtoINS.vcf.gz",manta,100000,100000,0.3,0.3,0.3,100000,true,true,0.3,0,0,cts,
+test2,"https://ftp-trace.ncbi.nlm.nih.gov/ReferenceSamples/giab/release/AshkenazimTrio/HG002_NA24385_son/NIST_SV_v0.6/GIAB_Evaluations/BoutrosLab-SV_curation_GIABv0.6/delly_GIABv0.6/norm_recall_HG002.merged__filt_DUPtoINS.vcf.gz",delly,100000,100000,0.3,0.3,0.3,100000,true,true,0.3,0,0,cts,
diff --git a/assets/samplesheet_small.csv b/assets/samplesheet_small.csv
@@ -0,0 +1,3 @@
+id,test_vcf,caller
+test1,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek/hg38/HG002.strelka.variants.chr21.vcf.gz,strelka
+test2,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek/hg38/HG002.bcftools.chr21.vcf.gz,bcftools
diff --git a/assets/samplesheet_somatic.csv b/assets/samplesheet_somatic.csv
diff --git a/assets/samplesheet_somatic_indel.csv b/assets/samplesheet_somatic_indel.csv
@@ -0,0 +1,3 @@
+id,test_vcf,caller,subsample
+test1,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek_full_test_somatic_v3.4.2/HCC1395T_vs_HCC1395N.freebayes.chr21.vcf.gz,freebayes,
+test2,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek_full_test_somatic_v3.4.2/HCC1395T_vs_HCC1395N.strelka.somatic_indels.vcf.gz,strelka,"TUMOR"
diff --git a/assets/samplesheet_somatic_snv.csv b/assets/samplesheet_somatic_snv.csv
@@ -0,0 +1,4 @@
+id,test_vcf,caller,subsample
+test1,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek_full_test_somatic_v3.4.2/HCC1395T_vs_HCC1395N.freebayes.chr21.vcf.gz,freebayes,
+test2,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek_full_test_somatic_v3.4.2/HCC1395T_vs_HCC1395N.manta.somatic_sv.vcf.gz,manta,"HCC1395_HCC1395T"
+test3,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek_full_test_somatic_v3.4.2/HCC1395T_vs_HCC1395N.strelka.somatic_snvs.vcf.gz,strelka,"TUMOR"
diff --git a/assets/samplesheet_somatic_sv.csv b/assets/samplesheet_somatic_sv.csv
@@ -0,0 +1,3 @@
+id,test_vcf,caller,subsample
+test1,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek_full_test_somatic_v3.4.2/HCC1395T_vs_HCC1395N.tiddit_sv_merge.vcf.gz,tiddit,"HCC1395_HCC1395T"
+test2,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/sarek_full_test_somatic_v3.4.2/HCC1395T_vs_HCC1395N.manta.somatic_sv.vcf.gz,manta,"HCC1395_HCC1395T"
diff --git a/assets/samplesheet_sv.csv b/assets/samplesheet_sv.csv
@@ -0,0 +1,4 @@
+id,test_vcf,caller,pctsize,pctseq,pctovl,refdist,chunksize,normshift,normdist,normsizediff,maxdist,typeignore,evaluationmode
+test1,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/SV_testdata/hg38/test/manta.HG002.chr21.vcf.gz,manta,0.3,0,0,100000,100000,0.3,0.3,0.3,100000,true,cts
+test2,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/SV_testdata/hg38/test/Ashkenazim_HG002.filtered.sv.chr21.vcf.gz,merged,0.3,0,0,100000,100000,0.3,0.3,0.3,100000,true,cts
+test3,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/SV_testdata/hg38/test/HG002_DRAGEN_SV_hg19.chr21.vcf.gz,dragen,0.3,0,0,100000,100000,0.3,0.3,0.3,100000,true,cts
diff --git a/assets/samplesheet_sv_hg37.csv b/assets/samplesheet_sv_hg37.csv
@@ -0,0 +1,3 @@
+id,test_vcf,caller
+test1,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/SV_testdata/hg37/test/HG002_delly_SV_hg19.chr21.vcf.gz,delly
+test2,https://raw.githubusercontent.com/kubranarci/benchmark_datasets/main/SV_testdata/hg37/test/HG002_manta_SV_hg19_genotype.chr21.vcf.gz,manta
diff --git a/assets/schema_input.json b/assets/schema_input.json
@@ -31,15 +31,6 @@
                 "minLength": 1,
                 "errorMessage": "Variant caller has to be defined. Can also be unknown, undefined or merged"
             },
-            "vartype": {
-                "type": "string",
-                "pattern": "^\\S+$",
-                "description": "Variant type to apply benchmarking",
-                "meta": ["vartype"],
-                "minLength": 1,
-                "errorMessage": "Variant type can be only one of these: small, sv, snv, indel and cnv",
-                "enum": ["small", "sv", "snv", "indel", "cnv"]
-            },
             "subsample": {
                 "type": "string",
                 "pattern": "^\\S+$",
@@ -165,6 +156,6 @@
                 "default": null
             }
         },
-        "required": ["test_vcf", "caller", "vartype", "id"]
+        "required": ["test_vcf", "caller", "id"]
     }
 }
diff --git a/conf/igenomes.config b/conf/igenomes.config
@@ -37,6 +37,7 @@ params {
             mito_name   = "chrM"
             macs_gsize  = "2.7e9"
             blacklist   = "${projectDir}/assets/blacklists/hg38-blacklist.bed"
+
         }
         'CHM13' {
             fasta       = "${params.igenomes_base}/Homo_sapiens/UCSC/CHM13/Sequence/WholeGenomeFasta/genome.fa"