remove ligand and alt res within chain by NatureGeorge · Pull Request #198 · steineggerlab/foldseek

NatureGeorge · 2023-10-24T04:59:45Z

solve #189

Cases to Test

PDB Entries with Alternative Residue Conformation

3JQH
1EJG

PDB Entries with Amino Acid Ligands

3LSW
3L1L

Something Should be Taken Care of

These changes assume the read-in PDB file contains enough information for gemmi to detect the entity type.

martin-steinegger · 2023-10-26T18:45:41Z

Thank you so much for the PR. I agree that we should remove ligands but why do you remove HET atoms?

NatureGeorge · 2023-10-27T12:19:34Z

Thank you so much for the PR. I agree that we should remove ligands but why do you remove HET atoms?

I did not make code changes on removing HET atoms. Do you refer to the following lines of code?

foldseek/src/strucclustutils/GemmiWrapper.cpp

Lines 284 to 286 in 8e68e86

    
           bool isHetAtomInList = res.het_flag == 'H' && threeAA2oneAA.find(res.name) != threeAA2oneAA.end(); 
        
           if (isHetAtomInList == false && res.het_flag != 'A') 
        
               continue;

These lines are the original code. (in other words, not committed by me)

From my understanding, removing HET atoms is within the logic of the original code, particularly when it meets residue type outside the alphabet of the threeAA2oneAA table, which contains the 20 standard amino acids and some kinds of modified residues.

foldseek/src/strucclustutils/GemmiWrapper.cpp

Lines 13 to 36 in 8e68e86

    
           GemmiWrapper::GemmiWrapper(){ 
        
               threeAA2oneAA = {{"ALA",'A'},  {"ARG",'R'},  {"ASN",'N'}, {"ASP",'D'}, 
        
                                {"CYS",'C'},  {"GLN",'Q'},  {"GLU",'E'}, {"GLY",'G'}, 
        
                                {"HIS",'H'},  {"ILE",'I'},  {"LEU",'L'}, {"LYS",'K'}, 
        
                                {"MET",'M'},  {"PHE",'F'},  {"PRO",'P'}, {"SER",'S'}, 
        
                                {"THR",'T'},  {"TRP",'W'},  {"TYR",'Y'}, {"VAL",'V'}, 
        
                                // modified res 
        
                                {"MSE",'M'}, {"MLY",'K'}, {"FME",'M'}, {"HYP",'P'}, 
        
                                {"TPO",'T'}, {"CSO",'C'}, {"SEP",'S'}, {"M3L",'K'}, 
        
                                {"HSK",'H'}, {"SAC",'S'}, {"PCA",'E'}, {"DAL",'A'}, 
        
                                {"CME",'C'}, {"CSD",'C'}, {"OCS",'C'}, {"DPR",'P'}, 
        
                                {"B3K",'K'}, {"ALY",'K'}, {"YCM",'C'}, {"MLZ",'K'}, 
        
                                {"4BF",'Y'}, {"KCX",'K'}, {"B3E",'E'}, {"B3D",'D'}, 
        
                                {"HZP",'P'}, {"CSX",'C'}, {"BAL",'A'}, {"HIC",'H'}, 
        
                                {"DBZ",'A'}, {"DCY",'C'}, {"DVA",'V'}, {"NLE",'L'}, 
        
                                {"SMC",'C'}, {"AGM",'R'}, {"B3A",'A'}, {"DAS",'D'}, 
        
                                {"DLY",'K'}, {"DSN",'S'}, {"DTH",'T'}, {"GL3",'G'}, 
        
                                {"HY3",'P'}, {"LLP",'K'}, {"MGN",'Q'}, {"MHS",'H'}, 
        
                                {"TRQ",'W'}, {"B3Y",'Y'}, {"PHI",'F'}, {"PTR",'Y'}, 
        
                                {"TYS",'Y'}, {"IAS",'D'}, {"GPL",'K'}, {"KYN",'W'}, 
        
                                {"CSD",'C'}, {"SEC",'C'}, 
        
                                // unknown 
        
                                {"UNK",'X'}}; 
        
           }

If the residue type is contained by the threeAA2oneAA table, the residue is kept and perform some filtering (NOTE: original code):

foldseek/src/strucclustutils/GemmiWrapper.cpp

Lines 287 to 291 in 8e68e86

    
           if (isHetAtomInList) { 
        
               bool hasCA = false; 
        
               for (gemmi::Atom &atom : res.atoms) { 
        
                   if (atom.name == "CA") { 
        
                       hasCA = true;

What I commit is the following lines of code (from line 288 to line 291):

foldseek/src/strucclustutils/GemmiWrapper.cpp

Lines 287 to 291 in c54dce5

    
           if (isHetAtomInList) { 
        
               bool notPolymer = res.entity_type != gemmi::EntityType::Polymer; 
        
               if (notPolymer == true) { 
        
                   continue; 
        
               }

The logic is that when it meets HET residue type that still falls within the threeAA2oneAA table, it has to check whether the residue is a member of the polymer chain, otherwise the residue is a non-polymer instance (i.e. ligand).

To implement this idea, I use gemmi's entity information. This requires the read-in PDB file to be standard enough for gemmi to determine the entity type of each residue. Perhaps there are some other ways to determine the entity type.

martin-steinegger · 2023-11-09T05:46:13Z

Thank you so much.

NatureGeorge added 2 commits October 24, 2023 12:36

remove ligand and alt res within chain

d5b4dab

verify entity type inside isHetAtomInList

c54dce5

martin-steinegger merged commit 0c32b3b into steineggerlab:master Nov 9, 2023

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

remove ligand and alt res within chain#198

remove ligand and alt res within chain#198
martin-steinegger merged 2 commits into
steineggerlab:masterfrom
NatureGeorge:fix_alt_aa_seq

NatureGeorge commented Oct 24, 2023

Uh oh!

martin-steinegger commented Oct 26, 2023

Uh oh!

NatureGeorge commented Oct 27, 2023

Uh oh!

martin-steinegger commented Nov 9, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

NatureGeorge commented Oct 24, 2023

Cases to Test

PDB Entries with Alternative Residue Conformation

PDB Entries with Amino Acid Ligands

Something Should be Taken Care of

Uh oh!

martin-steinegger commented Oct 26, 2023

Uh oh!

NatureGeorge commented Oct 27, 2023

Uh oh!

martin-steinegger commented Nov 9, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants