MCScanX icon indicating copy to clipboard operation
MCScanX copied to clipboard

Nothing is correct

Open M7MD-ElHELALY opened this issue 7 months ago • 2 comments

Hello everyone

My name is Mohamed, I have an issue with MCScanX that seems alot of people also have. I tried to figure out what is the input files problem but I tried every possible way I saw here or on reddit but no hope at all So I thought that it will be helpful to upload my input files so you can get the issue and help me repair them. Because every time I run MCScanX it gives me ZERO collinearity. I added .txt to the files so I can upload them.

Image

Amh.blast.txt Amh.gff.txt

M7MD-ElHELALY avatar May 23 '25 10:05 M7MD-ElHELALY

@M7MD-ElHELALY If mcscanx has no results, and you are running this in terminal. It is likely you need to put .blast and .gff file inside MCScanX script folder without creating a separate folder to run it. Also make sure blast and off have the same gene ID and remove the suffix '.txt' for both input files.

If you interested, there is a helper for faster preparing input data and easier running MCScanX in Conda environment. Please find the link: https://github.com/zx0223winner/MCScanX_Assistant

zx0223winner avatar Jul 26 '25 04:07 zx0223winner

@M7MD-ElHELALY unsure if you've been able to resolve this on your own, but it looks like you use the same notation for both your genes and the scaffolds their on, both are "Amh#'. If you can change one of them I think your issue might be resolved.

bauerlev avatar Aug 13 '25 23:08 bauerlev