Open liufy11 opened 4 years ago
Hey, I am not sure if I follow. Eval is only for the data that was generated by the simulation method. Is that what you are doing? You ran the simulation for SVs and generated reads, ... , called SVs and now wanted to merge to evaluate this result? Thanks Fritz
That's right !
You mean that the count between the simulated truth set is different? That is expected since I back then encoded the TRA differently so they have two entries each. Otherwise please try to explain what is the problem. Thanks Fritz
I think the difference is due to that SVs detected by different Softwares was not merged completely. for example, 3 simulated SVs can been detected by both SyRI and Sniffle, but ony 2 SVs were merged, so the SV number in SyRI_Sni_right.vcf is 4 (2+1+1); and the right SV number in SyRI_Sni.eval is only 3; Is that possible ?
I have never run the evaluation to evaluate multiple callers simultaneously if that is what is happening. So it looks at the VCF file as one call set. Sorry either my brain is occupied with something else or I am still not fully following... If you like we could have a short call next week and discuss this. Thanks Fritz
I merge my VCF files using "SURVIVOR merge SyRI_Ass_Sni.path 1000 1 1 1 0 30 SyRI_Ass_Sni.vcf" and after evalueing SyRI_Ass_Sni.vcf using "SURVIVOR eval SyRI_Ass_Sni.vcf simulated.sort.bed 1000 SyRI_Ass_Sni 1>SyRI_Ass_Sni.eval" I get the right vcf file named "SyRI_Ass_Sni_right.vcf" and the evalue file named "SyRI_Ass_Sni.eval" but the number of right SV counted by the above two files is different
I count the right number by "wc -l SyRI_Ass_Sni_right.vcf" or add up all the right SV number wroted in SyRI_Ass_Sni.eval
So which one is correct on earth ?