Question: Does SQUID accept STAR `WithinBAM` output?

Kingsford-Group / squid

SQUID detects both fusion-gene and non-fusion-gene structural variations from RNA-seq data

BSD 3-Clause "New" or "Revised" License

40 stars 22 forks source link

Question: Does SQUID accept STAR `WithinBAM` output? #8

Closed fgvieira closed 6 years ago

fgvieira commented 6 years ago

Dear all,

I've been trying to get my input files for SQUID, but I am not sure about the input. As far as I can see, SQUID accepts two types of input:

STAR: separate alignment and chimeric BAM files
BWA: combined BAM file of both concordant and discordant alignments

However, STAR also allow for a combined output (--chimOutType WithinBAM, either HardClip or SoftClip). Is this format compatible with SQUID BWA input (--bwa option)? And should it be HardClip or SoftClip?

thanks,

Congm12 commented 6 years ago

For the time being, it is recommended to separate chimeric reads and concordant reads for STAR. For chimeric split-read alignments, there are multiple SAM records corresponding to one paired-end read, however only one record is marked as chimeric. This may lead to erroneous or missing edges in SQUID's segment graph when the records are mixed with all other concordant SAM records, and finally reduces the accuracy. So when you are STAR available, it is recommended to NOT use WithinBam option.

fgvieira commented 6 years ago

ok, thanks!