HiSAT2 Alignment Rate Dropping with Cufflinks export option enabled

Question: HiSAT2 Alignment Rate Dropping with Cufflinks export option enabled

20 months ago by

I am trying to take trimmed and qc filtered paired RNA seq data and run it through HiSat2 > Stringtie > Cuffdiff for DE analysis. I am getting some interesting behavior with the HiSat2 alignment percentages. When the Report alignments tailored specifically for Cufflinks option is checked in galaxy or the --dta-cufflinks for command line users, the mapped percentage fall substantially on the same dataset. Does anyone have any idea what could be causing this?

For example with cufflinks option: 36198885 reads; of these: 36198885 (100.00%) were paired; of these: 17307719 (47.81%) aligned concordantly 0 times 15647243 (43.23%) aligned concordantly exactly 1 time 3243923 (8.96%) aligned concordantly >1 times

Same Dataset without: 36198885 reads; of these: 36198885 (100.00%) were paired; of these: 3167919 (8.75%) aligned concordantly 0 times 27036679 (74.69%) aligned concordantly exactly 1 time

5994287 (16.56%) aligned concordantly >1 times

rna-seq cufflinks hisat2 • 1.2k views

ADD COMMENT • link •

modified 13 months ago by Widmer, Giovanni • 150 • written 20 months ago by nick.andrews15 • 0

20 months ago by

Mo Heydarian ♦ 830

United States

Mo Heydarian ♦ 830 wrote:

Hello,

This is an interesting problem. The HISAT2 manual mentions a reduced mapping rate with the -dta option, but not how much to expect. I recommend reporting this issue to the HISAT2 github repository where the authors of the tool can comment or provide insight on the dramatic reduced mapping percentage you see with -dta-cufflinks option. The HISAT2 github repo can be found here: https://github.com/infphilo/hisat2/issues

Thanks for reporting your observation and for using Galaxy!

Cheers, Mo Heydarian

ADD COMMENT • link written 20 months ago by Mo Heydarian ♦ 830

19 months ago by

nick.andrews15 • 0

nick.andrews15 • 0 wrote:

Mo, Interestingly switching back to HiSAT 1 and using the -dta option for cufflinks yielded a roughly 85% alignment rate. I don't know what could be causing this, but I will use HiSAT for the time being I guess.

ADD COMMENT • link written 19 months ago by nick.andrews15 • 0

13 months ago by

Widmer, Giovanni • 150

US, Tufts University

Widmer, Giovanni • 150 wrote:

Hi, similar observation here. I align 100-nt single-end RNA-Seq reads to the built-in Sus scofa genome. I have 7,000,000 reads per sample. I get in the vicinity of 50% aligned sequences. What puzzles me is that if I BLAST randomly selected unaligned reads into genbank they all perfectly align to pig sequences. At most, I see 1-2 mismatches or an indel which make me wonder if HiSat struggles with reads that span a splice site. I don't think the issue has to do with FASTQ format because the alignment rate is slightly better if I use FASTQ formatted sequences as input as compared to the same input file in FASTA. I ran the sequences trough Trimmomatic and close to 100% passed. When I analyze the same files with TopHat, the % aligned is higher, ~75%.

thanks,

Giovanni Widmer

ADD COMMENT • link written 13 months ago by Widmer, Giovanni • 150

Please log in to add an answer.

5994287 (16.56%) aligned concordantly >1 times

Similar posts • Search »