AFmm settings: Run with ColabFold 1.5.5 with the following settings: AFmm v2.3 weights, 1 seed, 5 models, 5 recycles, local mmseqs2 server, unpaired_paired mode

BoltzGen yaml: entities:

Specify a designed protein chain

• protein: id: G

  random number between 40 and 80 of designed residues (inclusive)

  sequence: 40..80

Specification of the target which is extracted from a .cif file

• file: path: 2LGV_model1.pdb

  Which chain and residues in the .cif file to use as target (uses only chain A here)

  include: - chain: id: A

  Which regions of the target the design should or should NOT bind to

  Here we specify that the design should bind to residues 1..32 on chain A and not bind to residues 33..100 on chain A

  binding_types: - chain: id: A binding: 1..32 not_binding: 33..100

  Which regions of the target should have their structure specified

  Here we keep structure specified (visible) region 39..97 of chain A

  structure_groups: - group: visibility: 1 id: A res_index: 39..97

BoltzGen rescore settings: use_affinity=False, filter_bindingsite=False, filter_designfolding=False, refolding_rmsd_threshold=2.5, alpha=0, "design_ptm": 2, "neg_min_design_to_target_pae": 1, "design_to_target_iptm": 1, "plip_hbonds_refolded": 4, "plip_saltbridge_refolded": 4, "delta_sasa_refolded": 4, "neg_design_hydrophobicity": 7