Lab Assistant Tools
NextELN's AI Lab Assistant is more than chat and search — it's a computational wet-lab toolbox.
Paste a sequence, get a designed reagent, and save it straight into your notebook with the reasoning
and references attached. Open them from Tools in the app (/eln/tools).
Every tool carries a maturity label — see the Tool Catalog for the full status of each surface (Production / Beta / Research preview / Roadmap / Literature review).
🧊 3D Protein Structure Viewer · Beta
Paste a protein sequence and fold it with ESMFold, or enter a PDB ID or UniProt accession to pull an experimental or AlphaFold structure. Explore it interactively in 3D — cartoon, surface, or stick — with per-residue pLDDT confidence coloring for predicted models, and download the PDB.
💉 Antigenic Peptide Finder (antibodies for mice & rabbits) · Beta
Paste a protein and get a ranked list of the best peptides to raise a polyclonal antibody. It combines the classic immunoinformatics scales — Parker hydrophilicity, Kolaskar–Tongaonkar antigenicity, Emini surface accessibility, Karplus–Schulz flexibility, Chou–Fasman turns — into one immunogenicity score, penalizes synthesis-hostile features, and gives KLH/BSA conjugation guidance (where to add a terminal cysteine). A retrospective benchmark (successful vs. failed antigen peptides) is in progress.
🦠 mRNA Vaccine Candidate Designer · Research preview
From a viral genome (or a protein antigen): find open reading frames and antigens, predict B-cell and T-cell (MHC-I / MHC-II) epitopes, assemble a multi-epitope construct (signal peptide, adjuvant, and the standard AAY / GPGPG / KK / EAAAK linkers), codon-optimize for human expression, and wrap it with 5′/3′ UTRs and a poly-A tail. The output is a computational candidate package for expert review, experimental planning, and preclinical validation — not a bench-ready vaccine. Going from predicted epitopes to an experimentally useful mRNA vaccine involves expression, HLA diversity, immunodominance, RNA stability, formulation, and delivery, none of which this tool settles.
Grounded in the literature
Every tool is anchored to a curated library of 500+ primary references (AI in biomedicine, protein structure, antibody & epitope design, mRNA vaccines, multi-omics) that the assistant can cite. The reference system records evidence type, peer-reviewed vs. preprint, and last-review date — not just keyword relevance. See the background and methods in the toolbox manuscript.
Runs on your hardware, or ours
The lightweight steps run instantly in the app. The model-backed steps — ESMFold folding, BepiPred B-cell epitopes, and NetMHCpan / NetMHCIIpan T-cell epitopes — run on self-hosted GPU models inside your own network, so nothing leaves your lab. Until that service is connected, the tools still work using transparent, clearly-labeled heuristics. Model licensing is tracked in our model registry before any inference is offered commercially.
These tools are decision-support: in-silico predictions are hypotheses that require wet-lab validation.
Research notes are prepared with the NextELN AI Lab Assistant and scientifically reviewed by the NextELN science team, Excellgen, Inc.