Code to take an antibody PDB file and split it into separate Fvs
- The code needs to retain the PDB header for downstream steps
- Code needs to apply symmetry operators (e.g. 1b0w, 2rhe)
- decided to do this first with pdbsymm
- Some truncated sequences won't score highly enough to be flagged as antibodies - just score over the aligned region.
- Crystal packing sometimes identified as a complex. e.g. 1shm_1
- Read the PDB file and split into chains
- For each chain, scan with a library of VH and VL domains to locate these. This is done by alignment and the standard interface positions are recorded.
- For each VH/VL domain, find the CoG
- For each VH/VL domain measure the CofG distance to all others. If within a cutoff, then this is a potential VH/VL pair.
- For each potential VH/VL pair (note that these can be VL/VL or VH/VL and may be within the same chain) check the standard interface residues (assigned from the alignment) and ensure that they are close together
- Add back HETATMs (other than water)
- We now have assigned VH/VL pairs, so check those against other chains to look for antibody/antigen interactions. Note that an 'antigen' may be a different antibody.
- Now check the HETATMs for HET antigens
- Output the VH/VL domains together with any contacted antigen