Rule added to create gene counts file (w/ only gene name).

Author: krabbani3

SegDupAnalysis.snakefile

@@ -34,7 +34,7 @@ pos=[]
 subs=["all", "high_ident"]
 
 
-localrules: all, AnnotateResolvedTandemDups, GetUniqueGencodeUnresolvedDupGenes,  IntersectGenesWithFullSDList, FullDupToBed12, FullDupToLinks, MakeWMBed, MaskFile, ConvertHMMCopyNumberToCollapsedDuplications, SortSedef, FilterSedef, CountMaskedSedef, RemoveSedefTooMasked, MakeSedefGraph, MakeSedefGraphTable, FilterByGraphClusters, FullDupToBed12, FiltDupToBed12, GetUniqueGencodeUnresolvedDupGenesCN, GetUniqueGencodeUnresolvedDupGenes, GetGencodeMulticopy, GetGencodeMappedInDup, GetSupportedMulticopy,FindResolvedDuplicatedGenes, Bed12ToBed6, CombineGenesWithCollapsedDups, CombineDuplicatedGenes, MinimapGeneModelBed, FilterGencodeBed12, FindGenesInResolvedDups, SelectOneIsoform, SplitSplicedAndSingleExon, AnnotateLowCoverageFlanks, UnionMasked,GetNamedFasta, SelectDups, SortDups, GetDepthOverDups, FilterLowDepthDups, GetFullGeneCountTable, AddCollapsedGenes, GetCombinedTable, SelectDupsOneIsoform, GetFinalMerged, DupsPerContig, GetAllMultiGenes, AnnotateHighIdentity, GetTotalMasked, AnnotateResolvedTandemDups, GeneCountFact, GetFullGeneCountTable, FilterMultiExonBed, MappedSamIdentityDups, RemoveOriginal, RemoveBams, MakeSedefIntv, HighestIdentPairs, SelectHighIdent, GetCollapseByRange, GetCollapsedMask, GetCN
+localrules: all, AnnotateResolvedTandemDups, GetUniqueGencodeUnresolvedDupGenes,  IntersectGenesWithFullSDList, FullDupToBed12, FullDupToLinks, MakeWMBed, MaskFile, ConvertHMMCopyNumberToCollapsedDuplications, SortSedef, FilterSedef, CountMaskedSedef, RemoveSedefTooMasked, MakeSedefGraph, MakeSedefGraphTable, FilterByGraphClusters, FullDupToBed12, FiltDupToBed12, GetUniqueGencodeUnresolvedDupGenesCN, GetUniqueGencodeUnresolvedDupGenes, GetGencodeMulticopy, GetGencodeMappedInDup, GetSupportedMulticopy,FindResolvedDuplicatedGenes, Bed12ToBed6, CombineGenesWithCollapsedDups, CombineDuplicatedGenes, MinimapGeneModelBed, FilterGencodeBed12, FindGenesInResolvedDups, SelectOneIsoform, SplitSplicedAndSingleExon, AnnotateLowCoverageFlanks, UnionMasked,GetNamedFasta, SelectDups, SortDups, GetDepthOverDups, FilterLowDepthDups, GetFullGeneCountTable, AddCollapsedGenes, GetCombinedTable, SelectDupsOneIsoform, GetFinalMerged, DupsPerContig, GetAllMultiGenes, AnnotateHighIdentity, GetTotalMasked, AnnotateResolvedTandemDups, GeneCountFact, GetGeneCountTableAbbreviatedNames, FilterMultiExonBed, MappedSamIdentityDups, RemoveOriginal, RemoveBams, MakeSedefIntv, HighestIdentPairs, SelectHighIdent, GetCollapseByRange, GetCollapsedMask, GetCN
 
 
 
@@ -64,7 +64,8 @@ rule all:
         comb_with_depth="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.dups.one_isoform.txt.combined.depth",
         fact="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.dups.one_isoform.txt.combined.depth.filt.fact",
         gene_count="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.dups.one_isoform.txt.combined.depth.filt.gene_count",
-        gene_count_2column="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.dups.one_isoform.txt.combined.depth.filt.gene_count_multi_single",            
+        gene_count_2column="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.dups.one_isoform.txt.combined.depth.filt.gene_count_multi_single",
+        gene_count_abbrvNames="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.dups.one_isoform.txt.combined.depth.filt.gene_count.abbrv_names",
         asmMask=expand("{asm}.count_masked", asm=["assembly.orig.fasta", "assembly.masked.fasta", "assembly.repeat_masked.fasta", "assembly.union_masked.fasta"]),
         uniqueDupGenes="gencode.mapped.bam.bed12.dups.unique",
         sedef_high_uniq="sedef_out/all/final.sorted.bed.uniq.high",        
@@ -1466,7 +1467,8 @@ rule AnnotateOriginal:
     output:
         annot="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.sam.bed.dups.annot_orig",
     params:
-        sd=SD
+        sd=SD,
+        grid_opts=config["grid_medium"] # KEON
     shell:"""
 {params.sd}/AnnotateOriginal.py {input.mappedsambeddups} > {output.annot}
 """
@@ -1681,6 +1683,19 @@ cat {input.depth_filt} | awk '{{ if (NR == 1) {{ print "gene\\tresolved\\tcollap
 cat {output.gene_count_2column} | awk '{{ if (NR == 1) {{ print "gene\\tcopies";}} else {{  print $1"\\t"$2+$3;}} }}' > {output.gene_count}
 """
 
+rule GetGeneCountTableAbbreviatedNames:
+    input:
+        gene_count="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.dups.one_isoform.txt.combined.depth.filt.gene_count",
+    output:
+        gene_count_abbrvNames="gencode.mapped.bam.bed12.multi_exon.fasta.named.mm2.dups.one_isoform.txt.combined.depth.filt.gene_count.abbrv_names",
+    params:
+        sd=SD
+    shell:"""
+awk 'BEGIN {{ FS = "[ |\t]"}} \
+NR == 1 {{print}} \
+NR > 1 {{print $1,"\t" $3}}' {input.gene_count} > {output.gene_count_abbrvNames}
+"""
+
 rule cramBam:
     input:
         bam=config['bam'],