Tophat2 Vs Hisat2

Question

[poojasethiya]

I have used Tophat2 and Hisat2 to align the RNASeq data. For Tophat2 I used Htseq-count to get read counts and calculate the TPM. While for Hisat2 I used StringTie, prepDE.py to read counts, to further calculated the TPM. But on comparing replicates, the correlation between Tophat2-TPM is higher than Hisat2. Although Hisat2 is now recommended over Tophat2, but due to poor correlation between replicates by Hisat2 I am confused which one to use for further analysis.

Attached is the link of correlation: https://www.dropbox.com/s/ixgfwo7ro4pwu42/Align.png?dl=0

[EagleEye]

Hi,

Pease do the quantification for HISAT2 aligned samples exactly as you have done it for Tophat2 (or the vice versa). This difference might not necessarily because of aligners, it could be because of the quantification methods.

[h.mon]

Everything is different between your two pipelines, how can you be so sure Hisat2 is to blame?

[poojasethiya]

I am not blaming that Hisat2 is not good. Look at my question carefully, I am just comparing two tools for my analysis and confused about their results so asking for the opinion.

[h.mon]

My opinion is the two pipelines are completely different and I don't know the cause of the differences in the results. My suggestion is change only one "variable" to make comparisons easy, or use a data set where you know the truth to benchmark the two completely different pipelines.