Research & Technical Notes
Research & technical notes
Methods, architecture decisions, and literature reviews behind NextELN. These documents explain our reasoning; they are not claims that every described capability is a validated product.
🧬 The Enzyme That Wouldn't Fold: A Closed-Loop Computational Protein-Design Story
Research previewHow we used AlphaFold, generative sequence design, and a validated in-silico solubility oracle to diagnose why a decade-stubborn enzyme wouldn't express — and to design candidate fixes before a single new clone was made.
🧫 De Novo Protein Design: From Physics to Deep Learning — a Review
Literature reviewA survey of de novo protein design across five eras — physics-based sequence selection, deep-network hallucination, inverse folding (ProteinMPNN), generative diffusion (RFdiffusion/Chroma), and protein language models — built from a 507-paper corpus, with the modern generate→design→validate pipeline and how we implement it.
🧬 AI for the Wet Lab: A Toolbox for Protein Structure, Antibody Antigen Design, and mRNA Vaccine Discovery
Research previewA reference-grounded computational toolbox (3D structure, antigenic-peptide finder, mRNA-vaccine designer) built on 542 primary references. Prepared by NextELN Lab Assistant, an Excellgen, Inc. product.
🧰 Bioinformatics Tools in the AI Era: A Landscape Survey and Build-vs-Serve Framework
Literature reviewA survey of 2,485 bioinformatics tool papers across 20 domains, and a framework for which to build in-notebook (CPU) vs offer as GPU services. Prepared by NextELN Lab Assistant, an Excellgen, Inc. product.
📄 Bispecific Antibodies from Hybridoma Engineering to AI-Assisted In-silico Design
Literature reviewHistory, therapeutic mechanisms, pre-AI engineering, and modern AI/ML-assisted bispecific antibody design.
📄 In-silico Nanobody Design and Development
Research previewAlgorithms, evidence, and a reproducible sequence-first VHH design pipeline.
⚡ GPU-Powered AI Tools on NextELN: A Self-Hosted Inference Tier for the Wet Lab
BetaHow NextELN delivers GPU model inference (ESMFold, BepiPred, NetMHCpan) to the bench via the bio-infer service and an operator-brokered service workflow. Prepared by NextELN Lab Assistant, an Excellgen, Inc. product.