AF3 _merge_features_from_multiple_chains函数解读

AlphaFold3 msa_pairing模块的_merge_features_from_multiple_chains 函数的作用是合并多个蛋白质链的特征,以便在 AlphaFold3 处理中多个蛋白质链时形成统一的输入特征。

源代码:

MSA_FEATURES = ('msa', 'msa_mask', 'deletion_matrix', 'deletion_matrix_int')

MSA_GAP_IDX = residue_constants.restypes_with_x_and_gap.index('-')
MSA_PAD_VALUES = {'msa_all_seq': MSA_GAP_IDX,
                  'msa_mask_all_seq': 1,
                  'deletion_matrix_all_seq': 0,
                  'deletion_matrix_int_all_seq': 0,
                  'msa': MSA_GAP_IDX,
                  'msa_mask': 1,
                  'deletion_matrix': 0,
                  'deletion_matrix_int': 0}

MSA_FEATURES = ('msa', 'msa_mask', 'deletion_matrix', 'deletion_matrix_int')
SEQ_FEATURES = ('residue_index', 'aatype', 'all_atom_positions',
                'all_atom_mask', 'seq_mask', 'between_segment_residues',
                'has_alt_locations', 'has_hetatoms', 'asym_id', 'entity_id',
                'sym_id', 'entity_mask', 'deletion_mean',
                'prediction_atom_mask',
                'literature_positions', 'atom_indices_to_group_indices',
                'rigid_group_default_frame')
TEMPLATE_FEATURES = ('template_aatype', 'template_all_atom_positions',
                     'template_all_atom_mask')
CHAIN_FEATURES = ('num

你可能感兴趣的:(人工智能,生物信息学,深度学习,python)