DaehwanKimLab / hisat2

Graph-based alignment (Hierarchical Graph FM index)
GNU General Public License v3.0
473 stars 116 forks source link

HISAT2 vs STAR #273

Open asilvestris84 opened 3 years ago

asilvestris84 commented 3 years ago

After many comparisons between HISAT2 and STAR, I have come to the conclusion that the approach, which is the main innovative and revolutionary aspect of HISAT2, i.e. the possibility of introducing SNPs into the genome index, at least for how it is structured now represents a limit more than an advantage in that, especially in repeated regions, it introduces ambiguity thus reducing (and not a little) the number of uniquely mapped reads. I don't know if by modulating the number of SNPs to be introduced in the index, for example, by restricting it only to the most common ones or limiting to substitutions only without insertions and deletions, this bias can be somehow overcome. However, it remains an extremely valid and in my opinion absolutely brilliant attempt to advance scientific methods for the analysis of biological variability between individuals. I would very much like the authors to come up with a paper in which they honestly discuss these comparisons and how to overcome the limitations of each approach in their opinion.