Hi there,

I'm comparing software for the detection of fusion genes; star, tophat-fusion and defuse.

I've read about the Bioconductor chimera package which would be a convenient way of evaluating the software on an equal footing as the outputs of each tool can be imported into and analysed by the chimera module.

I've been able to import defuse data with no issues but, so far, I've not had any luck with STAR (Chimeric.out.junction) (or Tophat). First it says there aren't enough supporting reads (even though it's the same sample that's been analysed as defuse) and when I lower the value to min.support=2 and run the import command it produced the following error message after about 20-30 mins:

tst<-importFusionData("star","Chimeric.out.no.MT",parallel=TRUE, org="hs", min.support=2)
Error: 17688 errors; first error:
  Error in .getOneSeqFromBSgenomeMultipleSequences(x, name, start, NA, width, : sequence 1 not found

For more information, use bplasterror(). To resume calculation, re-call
  the function and set the argument 'BPRESUME' to TRUE or wrap the
  previous call in bpresume().

First traceback:
  36: importFusionData("star", "Chimeric.out.no.MT", parallel = TRUE,
          org = "hs", min.support = 2)
  35: .starImport(filename, ...)
  34: bplapply(tmp.loc.counts, function(x, z, j, k) .starFset(x, z,
          j, k), z = grHs, j = chr.sym, k = org, BPPARAM = p)
  33: bplapply(tmp.loc.counts, function(x, z, j, k) .starFset(x, z,
          j, k), z = grHs, j = chr.sym, k = org, BPPARAM = p)
  32: mclapply(X = X, FUN = FUN, ..., mc.set.seed = BPPARAM$setSeed,
          mc.silent = !BPPARAM$verbose, mc.cores = bpworkers(BPPARAM),
          mc.cleanup = if (BPPARAM$cleanup) BPPARAM$cleanupSignal else FALSE)
  31: lapply(seq_len(cores), inner.do)
  30: lappl

The command I used was:

tst<-importFusionData("star","Chimeric.out.no.MT",parallel=TRUE, org="hs", min.support=2)

The initial STAR command used to generate the Chimeric.out.junction file was:

STAR --genomeDir ${REF_DIR}/star-index --readFilesIn /13617_3.no.multihit.filt.R2.fastq 13617_3.no.multihit.filt.R1.fastq  --runThreadN 6 --chimSegmentMin 15 --chimJunctionOverhangMin 15

As I'm testing chimera for the first time, I'm assuming I'm diong something basic wrong but was wondering if you could provide any advice?

The Chimeric.output.junction file is very large, would you recommend some initial filtering or processing before attempting to import it?

Thanks

Angela

Hi Angela,

It seems that some of the detected sequences involved in the fusion cannot be extracted from the human genome present in bioconductor.

Which genome version did you used for STAR mapping? It is important that chimera version of genome and the one used for the mapping are the same. In chimera at the present time hg19 is used.

The size of STAR output is bery big and I added the min support spanning reads to reduce the input size. Please  try with min.support=10 to start to define the area of the error . ThenI will be pleased if you could share your data to check the chimera input problem.

Cheers

Raf

Thanks very much for the response.

The genome.fa file I used to generate the STAR index was GRCh37 downloaded from NCBI which should be the same as hg19 shouldn't it? Currently I'm tied to this file as it's used by the group I am working in.

When I try to import the STAR Chimeric.out.junction file using default parameters, I get a message saying something like "not enough supporting reads". When I run an awk statement on the file though and count up the number of times a fusion is seen, many of them (>100) are represented by more than 10 rows in the file. Some of them have over 500 records referring to the same fusion.

I'm going to try running STAR on a difference sample and will test the output of that to see if I'm getting the same issue but I think I probably will if it's a reference sequence problem.

Hi

This perhaps is not directly but kind of related  to the issue been discussed here.

So, I have used STAR for mapping using hg19 genome reference and Gencode v19 for annotation with the aim of detecting chimeric  junctions. I tried importing the STAR output into Chimera package for further filtering. But I realised that the annotation used in Chimera is of UCSC genes.  Is there a way to change the annotation used in Chimera to Gencode?