GATK RNAseq variant calling from HISAT and Trinity aligners

Question

[cristiana.ij.marques]

Hi guys,

I want to perform variant calling on my RNAseq dataset (with GATK) and I got stuck at the Split'N'Trim pre-processing step, as I want to apply the same (or similar) pipeline to two sets of reads: (1) aligned to a reference genome (with HISAT2) (2) aligned to a reference transcriptome (using Trinity/Bowtie2)

In their RNAseq Best Practices it is suggested the usage of mapping quality reassignment, however, HISAT2 already reassigns this score from 255 to 60. In this context, do you know if it's really necessary to specify this parameter?

-rf ReassignOneMappingQuality \ -RMQF 255 \ -RMQT 60

How about Trinity alignments, do you have any suggestions on how to procceed?

Thanks :)

[Ron]

You can try using https://github.com/davidliwei/rnaseqmut (mutation calling from RNAseq)

[akfraikncsu]

Undergoing a similar issue and was wondering if you had any updates on this?

[prasundutta87]

ReassignOneMappingQuality of GATK assumes that you have used STAR to align your RNAseq reads. STAR assigns the score 255 for its uniquly mapping reads which by default in SAM format means 'mapping quality not available/assigned'. GATK will get confused with this kind of scoring scheme and hence you use ReassignOneMappingQuality to change all the scores to 60 from 255.

HISAT2 (v2.0.4) assigns mapping quality of 60 to its uniquely mapped reads. Please pay attention to the version of the program. I believe its the same for its current version too, but you can check that. With this scoring scheme, GATK will have no problems in processing the data and you can easily not perform the ReassignOneMappingQuality step.