Entering edit mode
6.7 years ago
novicebioinforesearcher
▴
70
I am not able to understand what is the major difference between hisat2 and tophat2 (principle/algorithm), from what i could understand, hisat2 is faster than tophat2.
can someone please explain which one is better and why?
Hisat2 is "OK" at best and is generally outperformed by STAR.
Yes, I have read it, tophat2 uses bowtie and seed extension where as hisat2 uses a global search and then multiple local search (correct me if I am wrong here). Based on the comparison I did on one of the dataset available in GEO, alignment with tophat yields better alignment rates than with hisat2, this was confusing for me because most of the published comparison suggests that Hisat2 yields better results. Just to be sure I also used STAR where, results from star and hisat2 were comparable. This made me ask the question here.
Higher != better.
You didn't use a truth set or valid benchmarking. If I write an aligner which aligns 100% of the reads, is that the one you are going to use?
Agree! higher does not mean its better!
At the end of the day you are generating hypothesis to experimentally verify. If TopHat gives you leads (genes) that you are able to verify experimentally then by all means use TopHat (it is still a published method, even though the developers are recommending that you use HISAT2) and document the settings you used in your paper.
True..it finally boils down to experimental validation.