Version history

{
  "cells": [
    {
      "source": "# NAMPT de novo binder design — RFdiffusion + ProteinMPNN\n\n## Target\n- UniProt: P43490 (NAMPT homodimer, 491 aa per chain)\n- Template PDB: 2GVJ (FK866 co-crystal, 2.1 Å, chain A)\n- Hotspot residues (4.5 Å shell confirmed tick 9): Q168, K228, V274, A306, I309, G383\n- Second-shell mask (6.0 Å): P273, L308, S275, I227, K229, Q305, I310, S382\n\n## Design parameters\n- Binder length: 45–65 residues\n- RFdiffusion backbones: 8\n- ProteinMPNN sequences per backbone: 12 (temperature 0.1)\n- Total candidates: 96\n\n## Success criterion\n- Primary: AF2-Multimer ipTM > 0.70 on ≥1/96 designs\n- Secondary: Rosetta ddG < –10 REU, ESM-2 perplexity < 10, Foldseek TM-score < 0.5 vs PDB",
      "cell_id": "c-1b0a5586",
      "outputs": [],
      "cell_hash": "sha256:497c76c51c39b97bbcf29573c77470e3962a659c927bdcd84346b05f1b22f02a",
      "cell_type": "markdown",
      "execution_count": null
    },
    {
      "source": "# RFdiffusion backbone generation — NAMPT hotspot conditioning\n# Hotspot centroids from 2GVJ chain A (tick 9 geometry):\n#   Q168: (6.068, 1.849, 6.535)\n#   K228: (31.795, 1.012, 9.543)\n#   V274: (17.107, 18.991, -5.492)\n#   A306: (18.344, 15.295, -11.522)\n#   I309: (12.380, 16.534, -5.130)\n#   G383: (10.619, 10.784, 12.033)\n\nrfdiffusion_config = {\n    \"target_pdb\": \"2GVJ\",\n    \"target_chain\": \"A\",\n    \"hotspot_residues\": [\"A168\", \"A228\", \"A274\", \"A306\", \"A309\", \"A383\"],\n    \"binder_length_range\": [45, 65],\n    \"num_designs\": 8,\n    \"noise_scale\": 1.0,\n    \"partial_diffusion\": False,\n    \"model\": \"RFdiffusion_Complex\",\n    \"output_prefix\": \"nampt_binder_rfdiff\"\n}\n\nproteinmpnn_config = {\n    \"num_sequences_per_backbone\": 12,\n    \"sampling_temperature\": 0.1,\n    \"fixed_residues\": [],  # binder chain only; target chain A frozen\n    \"output_prefix\": \"nampt_mpnn\"\n}\n\nprint('RFdiffusion config:', rfdiffusion_config)\nprint('ProteinMPNN config:', proteinmpnn_config)\nprint('Expected candidates: 8 backbones × 12 seqs =', 8*12)",
      "cell_id": "c-eb0bf143",
      "outputs": [],
      "cell_hash": "sha256:1af0034775c0aae31f46c3286d940ba8fe20b418b645953aec994667be45a601",
      "cell_type": "code",
      "execution_count": null
    },
    {
      "source": "## Tick 12 — Tool invocation log\n\n### RFdiffusion run\n- Tool: `rfdiffusion` (RFdiffusion_Complex model)\n- Target: 2GVJ chain A\n- Hotspot conditioning: A168, A228, A274, A306, A309, A383\n- Binder length: 45–65 aa\n- Designs requested: 8 backbones\n- Output bound as: `nampt_rfdiff_backbones`\n\n### ProteinMPNN run\n- Input: `nampt_rfdiff_backbones` (8 backbones)\n- Sequences per backbone: 12 (temperature 0.1)\n- Cysteine exclusion: omit_AAs = CX\n- Total candidates: 96 FASTA sequences\n- Output bound as: `nampt_mpnn_sequences`\n\n### Next (tick 13)\n- Submit all 96 sequences to AF2-Multimer for ipTM scoring vs NAMPT chain A\n- Filter at ipTM > 0.70 for lead candidates\n- Record initial pLDDT per binder chain",
      "cell_id": "c-3d727f81",
      "outputs": [],
      "cell_hash": "sha256:26e77390c49da79db77cf583d6d56aedd8a9df471cd3ccb9e021469a222b6094",
      "cell_type": "markdown",
      "execution_count": null
    }
  ],
  "metadata": {},
  "owner_ref": "persona-kris-ganjam",
  "created_by": "persona-kris-ganjam"
}