You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Getting ResidueIdentifier without aligned SEQRES #456
Creating a new contact.ResidueIdentifier (not to be confused with ResidueNumber) requires that the Group have a valid alignedResIndex in the corresponding Compound. Is there a reasonable way to handle missing seqres (e.g. if the structure came from a default AtomCache (see #455))?
Would it be better to do contact matching by ResidueNumber instead of seqres index?
Contact matching based on seqres indices is a way to be able to compare residues in different chains of same entity when their numbering is inconsistent between chains. A good example of this is 3ddo where there's one entity with 6 chains. Each of those chains uses a different numbering: chain A starting at 1001, B at 2001, C at 3001 and so on... These cases can only be covered by doing residue comparisons based on seqres indices.
See test TestInterfaceClustering.test3DDO() that checks this.
It seems like the problem of aligning structures comes up often enough that we should have some utilities for handling this in a flexible way (e.g. use SEQRES if possible, otherwise default to a Smith-Waterman).
Agree @sbliven , I think we should work on making this more general, comparing 2 sets of contacts via an alignment or even in some topologically independent way.
This is something for after the 5.0 release. I'll move it to the next milestone.
Activity
josemduarte commentedon Apr 21, 2016
Contact matching based on seqres indices is a way to be able to compare residues in different chains of same entity when their numbering is inconsistent between chains. A good example of this is 3ddo where there's one entity with 6 chains. Each of those chains uses a different numbering: chain A starting at 1001, B at 2001, C at 3001 and so on... These cases can only be covered by doing residue comparisons based on seqres indices.
See test
TestInterfaceClustering.test3DDO()that checks this.sbliven commentedon Apr 22, 2016
It seems like the problem of aligning structures comes up often enough that we should have some utilities for handling this in a flexible way (e.g. use SEQRES if possible, otherwise default to a Smith-Waterman).
josemduarte commentedon Mar 7, 2018
Agree @sbliven , I think we should work on making this more general, comparing 2 sets of contacts via an alignment or even in some topologically independent way.
This is something for after the 5.0 release. I'll move it to the next milestone.
josemduarte commentedon Mar 11, 2021
I'm going to try to work on this for the 6.0.0 release. It requires some API breaking changes.