R/filter_variants.R

In bmannakee/smartcallr:

#' Given a data frame of variants including their pass
#' or fail codes as generated by MuTect2 or Oncotator
#' return a data frame filtered to contain only those variants
#' that failed ONLY because the log odds were too low.
#' TODO: parameters

filter_oncotated_variants <- function(var_fr){
  var_fr <- var_fr %>% dplyr::filter(!stringr::str_detect(Variant_Classification,'RNA')) %>% # Remove RNA calls
    mutate(all_filters=ifelse(germline_risk=="PASS" & t_lod_fstar=="PASS" & homologous_mapping_event=="PASS" &
                                alt_allele_in_normal=="PASS" & triallelic_site=="PASS" & clustered_events=="PASS" &
                                multi_event_alt_allele_in_normal=="PASS" & str_contraction=="PASS","PASS","FAIL")) %>%
    mutate(tlod_only=ifelse(germline_risk=="PASS" &  t_lod_fstar=="FAIL" & homologous_mapping_event=="PASS" &
                              alt_allele_in_normal=="PASS" & triallelic_site=="PASS" & clustered_events=="PASS" &
                              multi_event_alt_allele_in_normal=="PASS" & str_contraction=="PASS","PASS","FAIL")) %>%
    mutate('start'=Start_position,
           'end'=Start_position,
           'chr'=Chromosome,
           'ref'=Reference_Allele,
           'alt'=Tumor_Seq_Allele2,
           'tlod'=t_lod_fstar__INFO__,
           'alt_depth'=t_alt_count,
           'ref_depth'=t_ref_count) %>%
    dplyr::select(chr,start,end,ref,alt,tlod,Hugo_Symbol,Variant_Classification,Codon_Change,Protein_Change,all_filters,tlod_only,tumor_f,
                  alt_depth,ref_depth,germline_risk)
  var_fr
}