ProteinGPT is an AI‑powered web tool that lets users explore, visualize, and edit protein structures directly in the browser. It supports tasks such as mutating residues, removing water molecules, calculating pKa values, and extracting Mulliken charges from Gaussian log files, enabling researchers to quickly analyze computational chemistry data without leaving the platform. The interface also provides searchable access to UniProt entries and pre‑loaded examples like hemoglobin and EGFR for immediate experimentation.
Funding
Funding not disclosed
Founders
Product
Problem
Researchers in structural biology and computational chemistry often rely on fragmented tools and command-line workflows to visualize protein structures, edit residues, and extract data from simulation files, leading to time‑consuming and error‑prone processes.
Solution
ProteinGPT offers a web‑based, interactive platform that consolidates protein visualization, molecular editing, and computational chemistry analysis into a single interface. Users can load 3D structures, color by chain, mutate residues, remove solvent molecules, and calculate properties such as pKa values directly in the browser. The platform also supports uploading Gaussian log files to extract Mulliken charges and other quantum‑chemical data without leaving the web environment. By providing an intuitive UI with real‑time feedback, ProteinGPT streamlines structural biology workflows and reduces the need for multiple specialized software packages.
Target Audience
Primary users are structural biologists, computational chemists, and drug discovery scientists who need an integrated, browser‑based environment for protein modeling and analysis.
Features
- Interactive 3D protein viewer with multiple display modes (cartoon, stick, sphere) and chain‑based coloring
- In‑browser residue mutation tool allowing single‑point edits (e.g., mutate residue 64 to alanine)
- Automated removal of solvent molecules from loaded structures
- Built‑in pKa calculation for selected proteins
- Gaussian .log file parser that extracts Mulliken charges and other quantum‑chemical parameters
- Quick search for protein structures by UniProt ID (e.g., EGFR P00533) and PDB accession
- Chat‑style command interface for issuing commands and receiving immediate results