fix residue 3 to 1 mapping to cover nonstandard AAs too

Author: thomashopf

evcouplings/compare/pdb.py

@@ -20,7 +20,7 @@
 from Bio.PDB.MMCIF2Dict import MMCIF2Dict
 
 from evcouplings.utils.config import InvalidParameterError
-from evcouplings.utils.constants import AA3_to_AA1
+from evcouplings.utils.constants import AA3_to_AA1, AA3_to_AA1_FULL
 from evcouplings.utils.helpers import DefaultOrderedDict
 from evcouplings.utils.system import (
     valid_file, ResourceError, tempdir
@@ -788,7 +788,7 @@ def get_chain(self, chain, model=0, is_author_id=True):
             # (this should be unique and circumvents issues from 0 seqres values if selecting based on author chain ID)
             coord_id=lambda df: df.auth_seq_id.astype(str) + df.insertion_code,
             seqres_id=lambda df: df.label_seq_id.astype(str).replace("0", pd.NA).replace("", pd.NA),
-            one_letter_code=lambda df: df.label_comp_id.map(AA3_to_AA1, na_action="ignore"),
+            one_letter_code=lambda df: df.label_comp_id.map(AA3_to_AA1_FULL, na_action="ignore"),
             # note that MSE will now be labeled as HETATM, which was not the case with MMTF
             hetatm=lambda df: df.record_type == "HETATM",
         ).reset_index(

evcouplings/utils/constants.py

@@ -36,3 +36,163 @@
 AA3_to_AA1 = {
     v: k for k, v in AA1_to_AA3.items()
 }
+
+"""
+Mapping extracted from https://github.com/steineggerlab/foldseek/blob/8979d230fb64c7089380b652758d8705493ed4a5/src/strucclustutils/GemmiWrapper.cpp#L110
+with following Python code, after manually editing fall-through case:
+
+for row in AA_CODES.split("\n"):
+    row = row.strip()
+    if "return" not in row or row.startswith("//"):
+        continue
+
+    symbol = row.split('"')[1]
+    code = row.split("return ")[1].split(";")[0].replace("'", "")
+
+    if code not in code_to_symbol:
+        code_to_symbol[code] = []
+
+    code_to_symbol[code].append(symbol)
+    symbol_to_code[symbol] = code
+"""
+AA3_to_AA1_FULL = {
+    'ALA': 'A',
+    'ARG': 'R',
+    'ASN': 'N',
+    'ABA': 'A',
+    'ASP': 'D',
+    'ASX': 'B',
+    'CYS': 'C',
+    'CSH': 'S',
+    'GLN': 'Q',
+    'GLU': 'E',
+    'GLX': 'Z',
+    'GLY': 'G',
+    'HIS': 'H',
+    'ILE': 'I',
+    'LEU': 'L',
+    'LYS': 'K',
+    'MET': 'M',
+    'MSE': 'M',
+    'ORN': 'A',
+    'PHE': 'F',
+    'PRO': 'P',
+    'SER': 'S',
+    'THR': 'T',
+    'TRY': 'T',
+    'TRP': 'W',
+    'TYR': 'Y',
+    'UNK': 'X',
+    'VAL': 'V',
+    'SEC': 'C',
+    'PYL': 'O',
+    'SEP': 'S',
+    'TPO': 'T',
+    'PCA': 'E',
+    'CSO': 'C',
+    'PTR': 'Y',
+    'KCX': 'K',
+    'CSD': 'C',
+    'LLP': 'K',
+    'CME': 'C',
+    'MLY': 'K',
+    'DAL': 'A',
+    'TYS': 'Y',
+    'OCS': 'C',
+    'M3L': 'K',
+    'FME': 'M',
+    'ALY': 'K',
+    'HYP': 'P',
+    'CAS': 'C',
+    'CRO': 'T',
+    'CSX': 'C',
+    'DPR': 'P',
+    'DGL': 'E',
+    'DVA': 'V',
+    'CSS': 'C',
+    'DPN': 'F',
+    'DSN': 'S',
+    'DLE': 'L',
+    'HIC': 'H',
+    'NLE': 'L',
+    'MVA': 'V',
+    'MLZ': 'K',
+    'CR2': 'G',
+    'SAR': 'G',
+    'DAR': 'R',
+    'DLY': 'K',
+    'YCM': 'C',
+    'NRQ': 'M',
+    'CGU': 'E',
+    '0TD': 'D',
+    'MLE': 'L',
+    'DAS': 'D',
+    'DTR': 'W',
+    'CXM': 'M',
+    'TPQ': 'Y',
+    'DCY': 'C',
+    'DSG': 'N',
+    'DTY': 'Y',
+    'DHI': 'H',
+    'MEN': 'N',
+    'DTH': 'T',
+    'SAC': 'S',
+    'DGN': 'Q',
+    'AIB': 'A',
+    'SMC': 'C',
+    'IAS': 'D',
+    'CIR': 'R',
+    'BMT': 'T',
+    'DIL': 'I',
+    'FGA': 'E',
+    'PHI': 'F',
+    'CRQ': 'Q',
+    'SME': 'M',
+    'GHP': 'G',
+    'MHO': 'M',
+    'NEP': 'H',
+    'TRQ': 'W',
+    'TOX': 'W',
+    'ALC': 'A',
+    'SCH': 'C',
+    'MDO': 'A',
+    'MAA': 'A',
+    'GYS': 'S',
+    'MK8': 'L',
+    'CR8': 'H',
+    'KPI': 'K',
+    'SCY': 'C',
+    'DHA': 'S',
+    'OMY': 'Y',
+    'CAF': 'C',
+    '0AF': 'W',
+    'SNN': 'N',
+    'MHS': 'H',
+    'SNC': 'C',
+    'PHD': 'D',
+    'B3E': 'E',
+    'MEA': 'F',
+    'MED': 'M',
+    'OAS': 'S',
+    'GL3': 'G',
+    'FVA': 'V',
+    'PHL': 'F',
+    'CRF': 'T',
+    'BFD': 'D',
+    'MEQ': 'Q',
+    'DAB': 'A',
+    'AGM': 'R',
+    '4BF': 'Y',
+    'B3A': 'A',
+    'B3D': 'D',
+    'B3K': 'K',
+    'B3Y': 'Y',
+    'BAL': 'A',
+    'DBZ': 'A',
+    'GPL': 'K',
+    'HSK': 'H',
+    'HY3': 'P',
+    'HZP': 'P',
+    'KYN': 'W',
+    'MGN': 'Q'
+}