18 months to 8 weeks. Find your hit compound first.
Nanolix generates novel small-molecule candidates optimized for target binding affinity, ADMET profiles, and synthetic accessibility — exploring 10¹² chemical space computationally before you run a single assay.
Brute-force screening leaves most of chemical space unexamined
Pharma spends $2.6B per approved drug. High-throughput screening covers the corners of chemical space that are convenient to enumerate — not the regions where multi-property optima for your specific target are likely to sit. Lipinski-compliant compounds with acceptable ADMET profiles and synthesizable scaffolds occupy a narrow, non-contiguous region that enumeration mostly misses.
Hit identification is where early programs lose years and budget. A 50-person chemistry department can run iterative design cycles to work through this; a 5-person biotech team cannot. Generative methods change that — but only when synthesizability is a first-class constraint during generation, not a filter applied after the fact.
From target constraints to ranked candidates
Target Profile Input
You provide the target protein structure, binding site constraints, key residues, and property thresholds. We ingest PDB IDs or structural files directly. No reformatting required.
Generative Chemistry Exploration
Our generative models navigate a learned latent chemical space via gradient-guided sampling toward multi-property optima — not exhaustive enumeration. We explore where conventional screening doesn't look.
Multi-Property Optimization
Candidates are ranked by Pareto-optimal binding affinity, full 11-property ADMET panel, and synthetic accessibility score. You receive SDF output with predicted properties and synthesis routes — ready for CRO handoff.
Numbers from our methodology
Binding, ADMET, and synthesizability — jointly
Where Nanolix fits your program
Existing lead has off-target liabilities
Your current lead scaffold has hERG affinity or CYP3A4 inhibition that standard medicinal chemistry approaches haven't resolved after two design cycles.
We generate novel scaffolds with equivalent binding affinity to your target but explicit constraints excluding the liability flagged in your assays. Candidates ranked by predicted selectivity window.
Need IP-distinct analogs
A competitor has filed broad composition-of-matter claims around your lead scaffold class. You need novel candidates with equivalent efficacy profiles and clear structural distance.
Generative exploration produces diverse scaffolds with explicit diversity analysis relative to known IP space. You receive candidates confirmed synthesizable with standard CRO reagents.
Fragment hit needs growing
You have a confirmed fragment hit at ~200 Da with good ligand efficiency. Growing it to drug-like molecular weight while maintaining binding and physical properties is the challenge.
Fragment expansion with simultaneous ADMET modeling as molecular weight increases. Predicted solubility, permeability, and metabolic stability at each growth step. Top candidates with synthesis routes.
What researchers say
The candidates Nanolix returned weren't just structurally novel — the ADMET confidence intervals were honest. They flagged exactly where the model uncertainty was highest, which told us where to focus the first synthesis round. That transparency is not something you get from traditional virtual screening vendors.
We ran a scaffold-hopping engagement after two failed design cycles. The timeline was the thing — eight weeks to a prioritized candidate set with synthesis routes the CRO confirmed as feasible. That's months faster than what our internal capacity allowed. The binding prediction accuracy on retrospective validation was within what the team would accept.
Start with a target briefing
Send us your target. We map the binding site, run a sample generation against your constraints, and return candidate structures with predicted properties — before any contract. 30-minute call to align on parameters, sample output within 5 business days.
Request a Target Briefing