End-to-end KRAS G12C computational study

Shared replay ·

Open Propab
# Task: end-to-end computational study of KRAS G12C You are working as a computational biologist/chemist. Do real computation, not mock data. Every number you report must come from code you ran. If a step fails or a tool can't be installed, say so and use the closest working alternative. Don't fabricate outputs. Save all outputs in an organized directory (one folder per step) using standard domain file formats, so they can be opened directly in viewers. 1. **Sequence & evolution** - Fetch human KRAS, HRAS, NRAS + 10–15 RAS orthologs across species (UniProt). - MSA (MAFFT or Clustal Omega), per-residue conservation scores. - Phylogenetic tree (FastTree or IQ-TREE). - Outputs: .fasta, .aln, .sto, .nwk, conservation.csv 2. **Structure** - Download PDB 6OIM (KRAS G12C + sotorasib) and a wild-type GDP-bound KRAS. - Clean, add hydrogens, protonate at pH 7.4. - Superpose mutant vs wild-type, per-residue RMSD, identify the switch-II pocket. - Write conservation scores into the B-factor column. - Outputs: .pdb, .cif, .pqr, rmsd_per_residue.csv 3. **Cheminformatics** - ~30 molecules: sotorasib, adagrasib, other KRAS inhibitors from ChEMBL, + decoys. - RDKit: 3D conformers, descriptors (MW, logP, TPSA, HBD/HBA, QED), Lipinski, Morgan fingerprints, Tanimoto similarity matrix, scaffold clustering. - Outputs: .sdf, .mol2, .smi, .xyz, descriptors.csv, similarity_matrix.csv 4. **Docking** - Dock into the switch-II pocket (Vina or smina). Redock sotorasib first and report RMSD to the crystal pose. If > 2 Å, diagnose before continuing. - Outputs: complex + pose files (.pdbqt, .sdf, .pdb), scores.csv 5. **Molecular dynamics** - Short OpenMM run (explicit solvent, 1–5 ns, whatever hardware allows) on apo KRAS G12C and the sotorasib complex. - RMSD, RMSF, radius of gyration, ligand contacts over time. - Outputs: topology .pdb + .dcd/.xtc trajectory, analysis CSVs 6. **Quantum chemistry** - xtb, or psi4/pyscf with a small basis, on sotorasib's acrylamide warhead fragment: optimized geometry, HOMO/LUMO, ESP, partial charges. Comment on electrophilicity. - Outputs: .xyz, .molden, HOMO/LUMO/ESP .cube files 7. **Electrostatics** - APBS on the protein. - Outputs: .dx potential map 8. **Extra formats**: fetch or generate one real, relevant example of each: - cryo-EM map of a RAS complex (.mrc/.map from EMDB) - KRAS variants (.vcf from ClinVar/gnomAD), gene annotation (.gff3), region (.bed) - GenBank record for KRAS (.gb) - small-molecule crystal structure (.cif from COD) - RAS/MAPK pathway (.sbml) - mass spectrum (.mzML) or IR/NMR spectrum (.jdx) of a relevant compound 9. **Report** (markdown): methods with tool versions and parameters, key findings with numbers, sanity checks passed/failed, limitations (short MD, small basis, docking scoring caveats), next steps with more compute. Include a manifest listing every output file, its format, and what it contains.

Checking memory for prior KRAS context

Checking whether earlier sessions left context on this KRAS study or house conventions Establishing which sequence, structure, docking and MD tools are installed and which public hosts are reachable before starting

Reconnaissance is clean: MAFFT/FastTree are on PATH, network works, and the missing pieces (OpenMM, PySCF, docking, MDTraj) are installable. Starting the package installs in the background and fetching the RAS sequence set in parallel.

Collecting human KRAS/HRAS/NRAS plus 12 RAS orthologs across species for alignment and phylogenetic analysis

The install job holds the Python kernel, so I'll fetch the sequences through the bash lane in parallel.

Collecting human KRAS/HRAS/NRAS plus 12 RAS orthologs for alignment and phylogenetics while the package install runs

Retrieved the 15-sequence RAS family set from UniProt with accessions and lengths recorded

14/15 sequences retrieved; chasing the missing S. pombe ras1 with alternate query forms.

Completing the ortholog set with Schizosaccharomyces pombe Ras1

OpenMM/MDTraj/Vina/PySCF/PDB2PQR are installed. Now completing the ortholog set and running the alignment/conservation/phylogeny step.

Started Sep 27 02:37 PM54M tokens
This run was paused.
kras_study/04_docking/side_by_side_sotorasib_adagrasib.pdb
4,119 atoms · 170 residuesDrag to rotate · Scroll to zoom · Right-drag to pan
End-to-end KRAS G12C computational study · Shared replay