search

Rappsilber-Laboratory / AlphaLink2

AlphaLink2: Integrating crosslinking MS data into Uni-Fold-Multimer
Creative Commons Attribution 4.0 International
50 stars 16 forks source link

Index error #31

Open jimfeng9705 opened 3 months ago

jimfeng9705 commented 3 months ago

I encountered this error: IndexError: index 855 is out of bounds for axis 1 with size 805.

I attached the crosslink csv file. alphalink.csv

jimfeng9705 commented 3 months ago

The fasta file is attached here. fasta.txt

lhatsk commented 1 month ago

Sorry, this went under. Your chain A is only 657 residues long, how come they are crosslinks at residue position 933? There seems to be something wrong with your FASTA / crosslinking data alignment.

615 A 933 A 0.05

jimfeng9705 commented 1 month ago

Thanks for clarifying! I will try it again.

Two more questions:

  1. shall I use all crosslinks combined from various runs? Would more crosslinks help guide the modeling?
  2. The AlphaLink 2 models seem to have lower ptm and iptm scores than AlphaFold models. I tried to use higher recycle number such as 50. What else shall I try to do?

Get Outlook for iOShttps://aka.ms/o0ukef

From: lhatsk @.> Sent: Tuesday, October 8, 2024 3:31:29 AM To: Rappsilber-Laboratory/AlphaLink2 @.> Cc: Changjian Feng @.>; Author @.> Subject: Re: [Rappsilber-Laboratory/AlphaLink2] Index error (Issue #31)

[EXTERNAL]

Sorry, this went under. Your chain A is only 657 residues long, how come they are crosslinks at residue position 933? There seems to be something wrong with your FASTA / crosslinking data alignment.

615 A 933 A 0.05

— Reply to this email directly, view it on GitHubhttps://github.com/Rappsilber-Laboratory/AlphaLink2/issues/31#issuecomment-2399342406, or unsubscribehttps://github.com/notifications/unsubscribe-auth/BBIHPOOPLY3SCA4NT4AL4DTZ2OQ7DAVCNFSM6AAAAABNE273K6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOJZGM2DENBQGY. You are receiving this because you authored the thread.Message ID: @.***>

lhatsk commented 1 month ago
  1. shall I use all crosslinks combined from various runs? Would more crosslinks help guide the modeling?

I would use all crosslinks. In general, more crosslink information should help.

  1. The AlphaLink 2 models seem to have lower ptm and iptm scores than AlphaFold models. I tried to use higher recycle number such as 50. What else shall I try to do?

It's hard to say what's going on. How well are the crosslinks explained by the AlphaFold models? Is it a large target? Many chains? Disordered regions? I would go with more sampling instead of more recycling. You could also try v2 weights vs v3 weights.