Fusion Filtering ("other" reading frame ) edits (#553)

* edits as per issue #552

* add IGH@/IGL@

* Update analyses/fusion_filtering/run_fusion_merged.sh

add TODO

Co-Authored-By: Jaclyn Taroni <jaclyn.n.taroni@gmail.com>

Co-authored-by: Jaclyn Taroni <jaclyn.n.taroni@gmail.com>

analyses/fusion_filtering/04-project-specific-filtering.Rmd

@@ -12,11 +12,11 @@ params:
     input: string
   dataStranded:
     label: "Input filtered fusion dataframe"
-    value: scratch/standardFusionStrandedExp_QC_expression_GTExComparison_annotated.RDS
+    value: scratch/standardFusionStrandedExp_QC_expression_filtered_annotated.RDS
     input: file
   dataPolya:
     label: "Input filtered fusion dataframe"
-    value: scratch/standardFusionPolyaExp_QC_expression_GTExComparison_annotated.RDS
+    value: scratch/standardFusionPolyaExp_QC_expression_filtered_annotated.RDS
     input: file    
   numCaller:
     label: "Least Number of callers to have called fusion"
@@ -111,7 +111,6 @@ table(fusion_calls$Caller)
 
 # aggregate caller 
 fusion_caller.summary <- fusion_calls %>%
-  dplyr::filter(Fusion_Type != "other") %>%
   dplyr::select(Sample,FusionName,Caller,Fusion_Type) %>%
   group_by(FusionName, Sample ,Fusion_Type) %>%
   unique() %>%
@@ -122,17 +121,20 @@ fusion_caller.summary <- fusion_calls %>%
 fusion_calls <- fusion_calls %>%
   # remove local rearrangement/adjacent genes
   dplyr::filter(!grepl("LOCAL_REARRANGEMENT|LOCAL_INVERSION",annots))
-
+  
 #to add aggregated caller from fusion_caller.summary
 fusion_calls<-fusion_calls %>% 
-  dplyr::filter(Fusion_Type != "other") %>% dplyr::select(-Caller,-annots) %>%
+  dplyr::select(-Caller,-annots) %>%
   left_join(fusion_caller.summary,by=(c("Sample","FusionName","Fusion_Type"))) %>%
   dplyr::select(-JunctionReadCount,-SpanningFragCount,-Confidence,-LeftBreakpoint,-RightBreakpoint) %>% unique()
 
 #merge with histology file
 fusion_calls<-merge(fusion_calls,clinical,by.x="Sample",by.y="Kids_First_Biospecimen_ID")
 
-
+# filter for putative driver genes and mutifused genes per sample
+putative_driver_annotated_fusions <- fusion_calls %>%
+  dplyr::filter(!is.na(Gene1A_anno) | !is.na(Gene1B_anno) | !is.na(Gene2A_anno) | !is.na(Gene2B_anno)) %>%
+  unique()
 
 
 
@@ -141,7 +143,9 @@ fusion_calls<-merge(fusion_calls,clinical,by.x="Sample",by.y="Kids_First_Biospec
 
 
 ```{r}
-# Gene fusion should be in-frame
+# Gene fusion should be in-frame/frameshift 
+fusion_calls<-fusion_calls %>% 
+  dplyr::filter(Fusion_Type != "other")
 # AND
 #
 # 1. Called by at least n callers
@@ -237,15 +241,10 @@ QCGeneFiltered_recFusionUniq<-QCGeneFiltered_recFusion %>%
 
 
 ```{r}
-
-# filter for putative driver genes and mutifused genes per sample
-putative_driver_annotated_fusions <- fusion_calls %>%
-  dplyr::filter(!is.na(Gene1A_anno) | !is.na(Gene1B_anno) | !is.na(Gene2A_anno) | !is.na(Gene2B_anno)) %>%
-  unique()
   
 # merge putative annotated oncogenic and scavenged back non-oncogenic annotated, recurrent fusions
 putative_driver_fusions<-rbind(QCGeneFiltered_recFusionUniq,putative_driver_annotated_fusions) %>%
-  unique() %>% select (-broad_histology) %>%
+  unique() %>% dplyr::select (-broad_histology) %>%
   as.data.frame()
 
 write.table(putative_driver_fusions,file.path(outputfolder,"pbta-fusion-putative-oncogenic.tsv"),sep="\t",quote=FALSE,row.names = FALSE)

analyses/fusion_filtering/README.md

@@ -20,7 +20,7 @@ We also gather counts for recurrent fusions and fused genes found in more than 3
 * pbta-gene-expression-rsem-fpkm.stranded.rds : aggregated stranded fpm data
 
 #### Inputs used as reference
-* genelistreference.txt : known kinases, oncogenes, tumor suppressors, curated transcription factors [@doi:10.1016/j.cell.2018.01.029], COSMIC Cancer Gene Census list[https://cancer.sanger.ac.uk/census] . MYBL1 [@doi:10.1073/pnas.1300252110], SNCAIP [@doi:10.1038/nature11327], FOXR2 [@doi:10.1016/j.cell.2016.01.015], TTYH1 [@doi:10.1038/ng.2849], and TERT [@doi:10.1038/ng.3438; @doi:10.1002/gcc.22110; @doi:10.1016/j.canlet.2014.11.057; @doi:10.1007/s11910-017-0722-5] were added to the oncogene list and BCOR [@doi:10.1016/j.cell.2016.01.015] and QKI [@doi:10.1038/ng.3500] were added to the tumor suppressor gene list based on pediatric cancer literature review. 
+* genelistreference.txt : known kinases, oncogenes, tumor suppressors, curated transcription factors [@doi:10.1016/j.cell.2018.01.029], COSMIC Cancer Gene Census list[https://cancer.sanger.ac.uk/census] . MYBL1 [@doi:10.1073/pnas.1300252110], SNCAIP [@doi:10.1038/nature11327], FOXR2 [@doi:10.1016/j.cell.2016.01.015], TTYH1 [@doi:10.1038/ng.2849], and TERT [@doi:10.1038/ng.3438; @doi:10.1002/gcc.22110; @doi:10.1016/j.canlet.2014.11.057; @doi:10.1007/s11910-017-0722-5] were added to the oncogene list and BCOR [@doi:10.1016/j.cell.2016.01.015] and QKI [@doi:10.1038/ng.3500] were added to the tumor suppressor gene list based on pediatric cancer literature review. IGH-@,IGH@ , IGL-@ and IGL@ were also added to reference list as oncogenic genes because StarFusion output contains these gene symbols instead of IGL/IGH as per public databases.
 * fusionreference.txt :  known TCGA fusions
 * Brain_FPKM_hg38_matrix.txt.zip : GTex brain samples FPKM data
 The code to generate genelistreference.txt and fusionreference.txt is available here: https://gist.github.com/kgaonkar6/02b3fbcfeeddfa282a1cdf4803704794#file-format_reference_gene_list-r

analyses/fusion_filtering/references/genelistreference.txt

@@ -6752,3 +6752,7 @@
 "6751"	"GAK"	"PfamKinase"	"Kinase"
 "6752"	"PIK3C2B"	"PfamKinase"	"Kinase"
 "6753"	"HUNK"	"PfamKinase"	"Kinase"
+"6754"	"IGH@"	"addedToallOnco_Feb2017.tsv"	"Oncogene"
+"6755"	"IGH-@"	"addedToallOnco_Feb2017.tsv"	"Oncogene"
+"6756"	"IGL@"	"addedToallOnco_Feb2017.tsv"	"Oncogene"
+"6757"	"IGL-@"	"addedToallOnco_Feb2017.tsv"	"Oncogene"