I am working on Somatic mutations and am interested in identifying the somatic signatures in the given samples using “Somatic Signature “ package. I do not have the VCF files, but I have a data frame with reference allele and alternate allele. Using the package manual, I was successful in making the file and my file looks like this:


VRanges object with 90199 ranges and 0 metadata columns:


          seqnames                 ranges strand         ref              alt


             <Rle>              <IRanges>  <Rle> <character> <characterOrRle>


      [1]    chr17   [19347811, 19347811]      +           C                T


      [2]    chr19   [52219814, 52219814]      +           C                T


      [3]    chr17   [ 7674894,  7674894]      +           C                T


      [4]    chr17   [ 7674894,  7674894]      +           C                T


      [5]    chr17   [ 7674894,  7674894]      +           C                T


      ...      ...                    ...    ...         ...              ...


  [90195]     chr1 [227101200, 227101200]      +           G                C


  [90196]    chr10 [ 48409888,  48409888]      +           C                A


  [90197]    chr23 [135904469, 135904469]      +           G                A


  [90198]    chr12 [ 55687966,  55687966]      +           G                A


  [90199]    chr23 [ 19395196,  19395196]      +           T                A

However, when I apply mutationContext function, it gives me following error:
Error in mutationContext(sca_vr, BSgenome.Hsapiens.UCSC.hg19, unify = TRUE) : 
  Only SNV substitutions are currently supported.

I have been trying to work around it. Tried to add meta columns with my sca_vr object but it didn’t help. I do not have a reference sequence. But I am working with human genome so BSgenome should work with this. Can you suggest what can I do to make this work??

 

It look as if some of your variants are not single nucleotide variant substitutions (e.g. A>G) and other types of variants are not covered in the current release version of the software. You can check if other types of variants are present in your input data with

idx_snv = ref(vr) %in% DNA_BASES & alt(vr) %in% DNA_BASES
vr[!idx_snv]

where vr is your VRanges object as you already have it. Unsupported variants are for example ones where the reference or the alternative alleles are not single bases, but rather indels. You can also remove these variants in your analysis, and just continue with the SNVs:

vr_new = vr[idx_snv]