Residue Names

@functools.lru_cache(maxsize=64)
def letters_three_to_one(restype: str, *, default: str) -> str

from alphafold3.constants.residue_names import letters_three_to_one

# Standard amino acids
print(letters_three_to_one('ARG', default='X'))  # Returns: 'R'
print(letters_three_to_one('MSE', default='X'))  # Returns: 'M'
print(letters_three_to_one('ZZZ', default='X'))  # Returns: 'X'

CCD_NAME_TO_ONE_LETTER: Mapping[str, str]

from alphafold3.constants.residue_names import CCD_NAME_TO_ONE_LETTER

print(CCD_NAME_TO_ONE_LETTER['ALA'])  # 'A'
print(CCD_NAME_TO_ONE_LETTER['MSE'])  # 'M' (selenomethionine)
print(CCD_NAME_TO_ONE_LETTER['PHE'])  # 'F'

ALA = sys.intern('ALA')  # Alanine
ARG = sys.intern('ARG')  # Arginine
ASN = sys.intern('ASN')  # Asparagine
ASP = sys.intern('ASP')  # Aspartic acid
CYS = sys.intern('CYS')  # Cysteine
GLN = sys.intern('GLN')  # Glutamine
GLU = sys.intern('GLU')  # Glutamic acid
GLY = sys.intern('GLY')  # Glycine
HIS = sys.intern('HIS')  # Histidine
ILE = sys.intern('ILE')  # Isoleucine
LEU = sys.intern('LEU')  # Leucine
LYS = sys.intern('LYS')  # Lysine
MET = sys.intern('MET')  # Methionine
PHE = sys.intern('PHE')  # Phenylalanine
PRO = sys.intern('PRO')  # Proline
SER = sys.intern('SER')  # Serine
THR = sys.intern('THR')  # Threonine
TRP = sys.intern('TRP')  # Tryptophan
TYR = sys.intern('TYR')  # Tyrosine
VAL = sys.intern('VAL')  # Valine

UNK = sys.intern('UNK')  # Unknown amino acid
GAP = sys.intern('-')    # Gap character
UNL = sys.intern('UNL')  # Unknown ligand
MSE = sys.intern('MSE')  # Selenomethionine (non-standard but common in PDB)

A = sys.intern('A')  # Adenine
G = sys.intern('G')  # Guanine
C = sys.intern('C')  # Cytosine
U = sys.intern('U')  # Uracil

DA = sys.intern('DA')  # Deoxyadenosine
DG = sys.intern('DG')  # Deoxyguanosine
DC = sys.intern('DC')  # Deoxycytidine
DT = sys.intern('DT')  # Deoxythymidine
T = sys.intern('T')    # Thymine (single letter)

UNK_NUCLEIC_ONE_LETTER = sys.intern('N')   # Unknown nucleic acid
UNK_RNA = sys.intern('N')                  # Unknown RNA
UNK_DNA = sys.intern('DN')                 # Unknown DNA

POLYMER_TYPES_NUM = 29                              # len(POLYMER_TYPES)
POLYMER_TYPES_NUM_WITH_UNKNOWN = 30                 # len(POLYMER_TYPES_WITH_UNKNOWN)
POLYMER_TYPES_NUM_WITH_GAP = 29                     # len(POLYMER_TYPES_WITH_GAP)
POLYMER_TYPES_NUM_WITH_UNKNOWN_AND_GAP = 31         # len(POLYMER_TYPES_WITH_UNKNOWN_AND_GAP)
POLYMER_TYPES_NUM_ORDER_WITH_ALL_UNKS_AND_GAP = 32  # len(POLYMER_TYPES_WITH_ALL_UNKS_AND_GAP)

WATER_TYPES: tuple[str, ...] = ('HOH', 'DOD')

UNKNOWN_TYPES: tuple[str, ...] = (UNK, UNK_RNA, UNK_DNA, UNL)

from alphafold3.constants import residue_names

# Three-letter to one-letter
one_letter = residue_names.letters_three_to_one('ARG', default='X')
print(one_letter)  # 'R'

# Using mapping directly
from alphafold3.constants.residue_names import PROTEIN_COMMON_THREE_TO_ONE
print(PROTEIN_COMMON_THREE_TO_ONE['PHE'])  # 'F'

# One-letter to three-letter
from alphafold3.constants.residue_names import PROTEIN_COMMON_ONE_TO_THREE
print(PROTEIN_COMMON_ONE_TO_THREE['W'])  # 'TRP'

from alphafold3.constants.residue_names import (
    POLYMER_TYPES_ORDER,
    PROTEIN_TYPES,
    RNA_TYPES,
)

# Check if a residue is a standard protein type
if 'ALA' in PROTEIN_TYPES:
    print("ALA is a standard amino acid")

# Get the index of a polymer type
index = POLYMER_TYPES_ORDER['ALA']
print(f"ALA has index {index}")

# Iterate over RNA types
for rna_base in RNA_TYPES:
    print(f"RNA base: {rna_base}")

from alphafold3.constants.residue_names import (
    PROTEIN_TYPES_ONE_LETTER_TO_INT,
    RNA_TYPES_ONE_LETTER_WITH_UNKNOWN_TO_INT,
)

# Encode amino acid to integer
aa_code = PROTEIN_TYPES_ONE_LETTER_TO_INT['A']  # 0
print(f"Alanine code: {aa_code}")

# Encode RNA base to integer
rna_code = RNA_TYPES_ONE_LETTER_WITH_UNKNOWN_TO_INT['G']  # 1
print(f"Guanine code: {rna_code}")