test v0.1.2

pirl-unc · Aug 2, 2024 · 77a05f9 · 77a05f9
1 parent 6ea1935
commit 77a05f9
Show file tree

Hide file tree

Showing 5 changed files with 36 additions and 10 deletions.
diff --git a/src/nexuslib/pipelines/modules/svisionpro.nf b/src/nexuslib/pipelines/modules/svisionpro.nf
@@ -14,6 +14,8 @@ process runSVisionPro {
     input:
         tuple val(sample_id), path(tumor_bam_file), path(tumor_bam_bai_file), path(normal_bam_file), path(normal_bam_bai_file)
         path(reference_genome_fasta_file)
+        path(reference_genome_fasta_fai_file)
+        path(svisionpro_model_file)
         val(params_svisionpro)
         val(params_svisionpro_extract)
         val(output_dir)
@@ -23,21 +25,19 @@ process runSVisionPro {
 
     script:
         """
-        mkdir -p output/
+        mkdir -p ${sample_id}_svisionpro_outputs/
         SVision-pro \
             --target_path $tumor_bam_file \
             --base_path $normal_bam_file \
             --genome_path $reference_genome_fasta_file \
-            --model_path /SVision-pro-1.8/src/pre_process/model_liteunet_1024_8_16_32_32_32.pth \
-            --out_path output/ \
+            --model_path $svisionpro_model_file \
+            --out_path ${sample_id}_svisionpro_outputs/ \
             --sample_name $sample_id \
             --process_num ${task.cpus} \
             $params_svisionpro
-        mv output/* .
+        mv ${sample_id}_svisionpro_outputs/* .
         python /opt/SVision-pro/extract_op.py \
             --input_vcf ${sample_id}*.vcf \
             $params_svisionpro_extract
         """
 }
-
-
diff --git a/...ib/pipelines/variant_calling/long_read_dna_variant_calling_svisionpro/README.md b/...ib/pipelines/variant_calling/long_read_dna_variant_calling_svisionpro/README.md
@@ -22,7 +22,9 @@ nexus run --nf-workflow long_read_dna_variant_calling_svisionpro.nf \
     --samples_tsv_file SAMPLES_TSV_FILE \
     --output_dir OUTPUT_DIR \
     --reference_genome_fasta_file REFERENCE_GENOME_FASTA_FILE \
-    --params_svisionpro '"--detect_mode somatic"' \
+    --reference_genome_fasta_fai_file REFERENCE_GENOME_FASTA_FAI_FILE \
+    --svisionpro_model_file SVISIONPRO_MODEL_FILE \
+    --params_svisionpro '"--detect_mode somatic --preset hifi --min_supp 3 --min_mapq 20 --min_sv_size 30 --max_sv_size 1000000 --device cpu --img_size 256"' \
     --params_svisionpro_extract '"--extract somatic --min_supp 3"'
 ```
 
@@ -42,7 +44,9 @@ required arguments:
 
 optional arguments:
     --reference_genome_fasta_file       :   Reference genome FASTA file (default: /datastore/lbcfs/collaborations/pirl/seqdata/references/hg38.fa).
-    --params_svisionpro                 :   SVision-pro parameters (default: '"--detect_mode somatic"').
+    --reference_genome_fasta_fai_file   :   Reference genome FASTA.FAI file (default: /datastore/lbcfs/collaborations/pirl/seqdata/references/hg38.fa.fai).
+    --svisionpro_model_file             :   SVision-pro model file (default: /datastore/lbcfs/collaborations/pirl/seqdata/tool-resources/svisionpro/SVision-pro-1.8/src/pre_process/model_liteunet_256_8_16_32_32_32.pth).
+    --params_svisionpro                 :   SVision-pro parameters (default: '"--detect_mode somatic --preset hifi --min_supp 3 --min_mapq 20 --min_sv_size 30 --max_sv_size 1000000 --device cpu --img_size 256"').
                                             Note that the parameters need to be wrapped in quotes.
     --params_svisionpro_extract         :   SVision-pro extract_op.py parameters (default: '"--extract somatic --min_supp 3"').
                                             Note that the parameters need to be wrapped in quotes.
@@ -67,6 +71,12 @@ optional arguments:
 `--reference_genome_fasta_file`
 * Reference genome FASTA files can be found in /datastore/lbcfs/collaborations/pirl/seqdata/references/ on LBG.
 
+`--reference_genome_fasta_fai_file`
+* Reference genome FASTA.FAI files can be found in /datastore/lbcfs/collaborations/pirl/seqdata/references/ on LBG.
+
+`--svisionpro_model_file`
+* Refer to the [SVision-pro documentation](https://github.com/songbowang125/SVision-pro).
+
 `--params_svisionpro`
 * Refer to the [SVision-pro documentation](https://github.com/songbowang125/SVision-pro).
 * The following parameters for `SVision-pro` are already included in `nexus` module for `SVision-pro` and should not be specified:

diff --git a/...ling/long_read_dna_variant_calling_svisionpro/long_read_dna_variant_calling_svisionpro.nf b/...ling/long_read_dna_variant_calling_svisionpro/long_read_dna_variant_calling_svisionpro.nf
@@ -12,7 +12,9 @@ params.samples_tsv_file = ''
 params.output_dir = ''
 // Optional arguments
 params.reference_genome_fasta_file = '/datastore/lbcfs/collaborations/pirl/seqdata/references/hg38.fa'
-params.params_svisionpro = '--detect_mode somatic --preset hifi --min_supp 3 --min_mapq 20 --min_sv_size 30 --max_sv_size 1000000 --device cpu'
+params.reference_genome_fasta_fai_file = '/datastore/lbcfs/collaborations/pirl/seqdata/references/hg38.fa.fai'
+params.svisionpro_model_file = '/datastore/lbcfs/collaborations/pirl/seqdata/tool-resources/svisionpro/SVision-pro-1.8/src/pre_process/model_liteunet_256_8_16_32_32_32.pth'
+params.params_svisionpro = '--detect_mode somatic --preset hifi --min_supp 3 --min_mapq 20 --min_sv_size 30 --max_sv_size 1000000 --device cpu --img_size 256'
 params.params_svisionpro_extract = '--extract somatic --min_supp 3'
 params.delete_work_dir = false
 
@@ -50,7 +52,9 @@ if (params.help) {
 
     optional arguments:
         --reference_genome_fasta_file       :   Reference genome FASTA file (default: /datastore/lbcfs/collaborations/pirl/seqdata/references/hg38.fa).
-        --params_svisionpro                 :   SVision-pro parameters (default: '"--detect_mode somatic"').
+        --reference_genome_fasta_fai_file   :   Reference genome FASTA.FAI file (default: /datastore/lbcfs/collaborations/pirl/seqdata/references/hg38.fa.fai).
+        --svisionpro_model_file             :   SVision-pro model file (default: /datastore/lbcfs/collaborations/pirl/seqdata/tool-resources/svisionpro/SVision-pro-1.8/src/pre_process/model_liteunet_256_8_16_32_32_32.pth).
+        --params_svisionpro                 :   SVision-pro parameters (default: '"--detect_mode somatic --preset hifi --min_supp 3 --min_mapq 20 --min_sv_size 30 --max_sv_size 1000000 --device cpu --img_size 256"').
                                                 Note that the parameters need to be wrapped in quotes.
         --params_svisionpro_extract         :   SVision-pro extract_op.py parameters (default: '"--extract somatic --min_supp 3"').
                                                 Note that the parameters need to be wrapped in quotes.
@@ -62,6 +66,8 @@ if (params.help) {
         samples_tsv_file                    :   ${params.samples_tsv_file}
         output_dir                          :   ${params.output_dir}
         reference_genome_fasta_file         :   ${params.reference_genome_fasta_file}
+        reference_genome_fasta_fai_file     :   ${params.reference_genome_fasta_fai_file}
+        svisionpro_model_file               :   ${params.svisionpro_model_file}
         params_svisionpro                   :   ${params_svisionpro}
         params_svisionpro_extract           :   ${params_svisionpro_extract}
         delete_work_dir                     :   ${params.delete_work_dir}
@@ -85,6 +91,8 @@ workflow LONG_READ_DNA_VARIANT_CALLING_SVISIONPRO {
     take:
         input_bam_files_ch             // channel: [val(sample_id), path(tumor_bam_file), path(tumor_bam_bai_file), path(normal_bam_file), path(normal_bam_bai_file)]
         reference_genome_fasta_file
+        reference_genome_fasta_fai_file
+        svisionpro_model_file
         params_svisionpro
         params_svisionpro_extract
         output_dir
@@ -93,6 +101,8 @@ workflow LONG_READ_DNA_VARIANT_CALLING_SVISIONPRO {
         runSVisionPro(
             input_bam_files_ch,
             reference_genome_fasta_file,
+            reference_genome_fasta_fai_file,
+            svisionpro_model_file,
             params_svisionpro,
             params_svisionpro_extract,
             output_dir
@@ -103,6 +113,8 @@ workflow {
     LONG_READ_DNA_VARIANT_CALLING_SVISIONPRO(
         input_bam_files_ch,
         params.reference_genome_fasta_file,
+        params.reference_genome_fasta_fai_file,
+        params.svisionpro_model_file,
         params_svisionpro,
         params_svisionpro_extract,
         params.output_dir

diff --git a/test/data/indices/svisionpro/model_liteunet_256_8_16_32_32_32.pth b/test/data/indices/svisionpro/model_liteunet_256_8_16_32_32_32.pth
diff --git a/test/variant_calling/test_long_read_dna_variant_calling_svisionpro.py b/test/variant_calling/test_long_read_dna_variant_calling_svisionpro.py
@@ -11,6 +11,8 @@ def test_long_read_dna_variant_calling_svisionpro():
     normal_dna_bam_file = get_data_path(name='bam/hg38_tp53_normal_long_read_dna.bam')
     normal_dna_bam_bai_file = get_data_path(name='bam/hg38_tp53_normal_long_read_dna.bam.bai')
     reference_genome_fasta_file = get_data_path(name='fasta/hg38_chr17_1-8000000.fa')
+    reference_genome_fasta_fai_file = get_data_path(name='fasta/hg38_chr17_1-8000000.fa.fai')
+    svisionpro_model_file = get_data_path(name='indices/svisionpro/model_liteunet_256_8_16_32_32_32.pth')
     temp_dir = os.getcwd() + '/tmp'
     intermediate_dir = temp_dir + '/intermediate/test_long_read_dna_variant_calling_svisionpro'
     work_dir = temp_dir + '/work/test_long_read_dna_variant_calling_svisionpro'
@@ -33,6 +35,8 @@ def test_long_read_dna_variant_calling_svisionpro():
         '-w', work_dir,
         '--samples_tsv_file', intermediate_dir + '/samples.tsv',
         '--reference_genome_fasta_file', reference_genome_fasta_file,
+        '--reference_genome_fasta_fai_file', reference_genome_fasta_fai_file,
+        '--svisionpro_model_file', svisionpro_model_file,
         '--params_svisionpro', '"--detect_mode somatic --preset hifi --min_supp 3 --min_mapq 20 --min_sv_size 30 --max_sv_size 1000000 --device cpu"',
         '--params_svisionpro_extract', '"--extract somatic --min_supp 3"',
         '--output_dir', output_dir,