Sounds right. You can look at some reads that were mapped by STAR but not by tophat and see how many mismatches they have
If I used Tophat to align reads to human genome and I got ~72% alignment rate. Then I realigned the reads using STAR and around 90% of the reads were mapped. I also used the same commands of tophat to align another dataset and got 90% alignments rate. So I don't think I am not using tophat incorrectly.
I just used trimgalore before alignment . This confuses me. Can we have explanation?
tophat -p 4 --no-mixed -G genes.gtf --library-type fr-unstranded -o out genome reads_val_1.fq.gz reads_val_2.fq.gz
2 answers
Tophat is an inferior and antiquated tool. Its inferiority has been born out in countless comparisons. You should never use tophat2 for new projects.
I think also I am using the default parameters of Star which could be the reason for the difference. 10 mismatches in Star vs 2 only in tophat. Also the softclipping default is high in Star vs no in tophat.
Log in to answer this question.