gpertea / stringtie

Transcript assembly and quantification for RNA-Seq
MIT License
386 stars 78 forks source link

Why stringtie2.1.4 is not better than stringtie1.3.5 ? #305

Open jinghui1012 opened 4 years ago

jinghui1012 commented 4 years ago

I'm testing the performance of stringtie1.3.5 and stringtie2.1.4 and expect the performance of stringtie2.1.4 to be better than stringtie1.3.5. But the result is not as expected. I simulated two isoforms, the ratio of the number of molecules is 0.1:0.9, with flux-simulator for short reads of mRNA and ran stringtie2.1.4 and stringtie1.3.5 on this data (reads_sort.bam). The output file of stringtie1.3.5 and stringtie2.1.4 are stringtie1.3.5.gtf and stringtie2.1.4.gtf respectively. Summary files records the expression ratio of the two isoforms. That is, their expression ratio should be 0.1 and 0.9. https://drive.google.com/file/d/1WCbGvqm8ruxZWv6SCstJxkCetOYMmp00/view?usp=sharing