pahase

bin/split_samples.py

@@ -95,7 +95,7 @@ def split_samples(sample_list, batch_size):
 def write_sample_lists(sample_list_split, outdir):
     i = 1
     for lst in sample_list_split:
-        sample_file = outdir + "/samples_" + str(i) + ".txt"
+        sample_file = outdir + "/samples_" + str(i)
         with open(sample_file, 'w') as sf:
             sf.write(('\n').join(lst))
             sf.write('\n')

conf/base.config

@@ -39,6 +39,13 @@ process {
         memory = { check_max( 200.GB * task.attempt, 'memory' ) }
     }
 
+    withLabel:process_phase {
+        cpus   = { check_max( 8    * task.attempt, 'cpus'    ) }
+        memory = { check_max( 20.GB * task.attempt, 'memory' ) }
+        time   = { check_max( 12.h  * task.attempt, 'time'   ) }
+    }
+
+
     withLabel:error_ignore {
         errorStrategy = 'ignore'
     }

modules/local/split_samples/main.nf

@@ -7,7 +7,7 @@ process SPLIT_SAMPLES {
     path(vcf_in)
 
     output:
-    path('samples_*.txt'), emit: sample_lists
+    path('samples_*'), emit: sample_lists
 
     script:
     """

modules/local/split_vcfs/main.nf

@@ -1,15 +1,17 @@
 process SPLIT_VCFS {
 
+    label 'process_medium'
+
     input:
     tuple path(vcf), path(sample_list)
 
     output:
-    tuple path('vcf_sample_subset.vcf.gz'), path('vcf_sample_subset.vcf.gz.csi'), emit: split_vcfs
+    tuple path('vcf_sample_subset_*.vcf.gz'), path('vcf_sample_subset_*.vcf.gz.csi'), emit: split_vcfs
 
     script:
     """
-    bcftools view -S $sample_list $vcf -Oz -o vcf_sample_subset.vcf.gz
-    bcftools index vcf_sample_subset.vcf.gz
+    bcftools view -S $sample_list $vcf -Oz -o vcf_sample_subset_${sample_list}.vcf.gz
+    bcftools index vcf_sample_subset_${sample_list}.vcf.gz
     """
 
 }

modules/nf-core/glimpse2/phase/main.nf

@@ -1,6 +1,6 @@
 process GLIMPSE2_PHASE {
     tag "$meta.id"
-    label 'process_medium'
+    label 'process_phase'
 
     beforeScript  """
     if cat /proc/cpuinfo | grep avx2 -q

workflows/run_glimpse.nf

@@ -10,20 +10,31 @@ workflow RUN_GLIMPSE {
     vcf_samples = vcf.combine(SPLIT_SAMPLES.out.sample_lists.flatten())
 
     SPLIT_VCFS(vcf_samples)
+    // // SPLIT_VCFS.out.split_vcfs.view()
 
     ref = channel.fromPath("${params.refdir}*.bin")
+
+    phase_meta = [id: 'phase']
 
     phase_input = SPLIT_VCFS.out.split_vcfs.combine(ref).map{
                                                                     vcf, index , ref_bin ->
-                                                                    [[], vcf, index, [], [], [], ref_bin, [], []]
+                                                                    [phase_meta, vcf, index, [], [], [], ref_bin, [], []]
                                                                 }
     phase_input2 = ['', params.fasta, params.fai]
 
-    //phase_input.view()
-
-
     GLIMPSE2_PHASE(phase_input, phase_input2)
 
     GLIMPSE2_PHASE.out.versions.view()
 
+    // phase_input = vcf_samples.combine(ref).map {
+    //                                                 vcf, samples, ref_bin ->
+    //                                                 [phase_meta, vcf, [], samples, [], [], ref_bin, [], []]
+
+    //                                                 }
+
+    // phase_input2 = ['', params.fasta, params.fai]
+
+    // GLIMPSE2_PHASE(phase_input, phase_input2)
+
+
 }