Genentech
Same PDB Chain IDs
Hello. When running your ab model weights on an input CSV containing a uid column as well as both heavy and light chain sequence columns, it appears as though your code produced PDB files containing only a single unique PDB chain ID (i.e., A). If the predicted PDB does in fact contain both heavy and light chain structures in it (as a single complex), is it possibly to distinguish the heavy chain structure from the light chain structure within the same PDB file by e.g., giving the heavy chain a chain ID of A and the light chain a chain ID of B?
Related to this, how does EquiFold combine antibody sequence inputs? For example, if I provide heavy and light sequences as input, how does EquiFold modify the sequence inputs to link the two chains together in its structure prediction? For example, in the PDB structure shown below, how does EquiFold construct the link between chains shown here?
