Home Services Projects Software Resources Publications About Contact
  Core Capability

From Molecule to Medicine, Faster

Our computational drug discovery pipeline combines AI-powered virtual screening, property prediction, and generative molecular design to reduce the time and cost of bringing new therapeutics to market.

15+
Years Experience
30+
EU Research Projects
700K+
Compounds Screened
Top 2%
Stanford University Ranking

Our Drug Discovery Pipeline

A complete computational toolkit for every stage of the drug discovery journey

1

Target Identification

Identify and validate biological targets using computational methods, protein structure analysis, and literature-driven target prioritization.

Virtual Screening & Hit Identification

Screen millions of compounds in silico using structure-based (molecular docking with AutoDock Vina, RxDock) and ligand-based approaches to find promising hits.

3

ADMET & Toxicity Prediction

Predict absorption, distribution, metabolism, excretion, and toxicity properties early using validated AI/ML models — before costly synthesis.

De Novo Molecular Design

Generate novel, drug-like molecules using our hybrid reinforcement learning framework combining GPT-based generative models with multi-objective optimization and retrosynthetic feasibility analysis.

5

Lead Optimization

Multi-parameter optimization guided by machine learning to improve potency, selectivity, and drug-likeness of lead compounds.

Powered by Our Software

Purpose-built platforms for computational drug discovery

Enalos Asclepios KNIME Nodes

A user-friendly toolkit offering state-of-the-art functionalities for preparing bioinformatics and cheminformatics components in molecular docking and molecular dynamics simulations.

  • Zero-code philosophy
  • Seamless KNIME integration
  • Molecular docking (AutoDock Vina & RxDock)
  • Structure-based drug design workflows
  • Reproducible computational pipelines
Learn more

Isalos Analytics Platform

End-to-end workflow automation for data manipulation, machine learning model development, and predictive analytics — no programming required.

  • AutoML & Advanced Statistics
  • No-code environment
  • Comprehensive Design of Experiments suite
  • QSAR/QSPR model building
  • Data-driven decision making
Learn more

Generative AI for De Novo Drug Design

Designing novel molecules that don't yet exist in any chemical library

NovaMechanics has developed a hybrid reinforcement learning framework that combines a GPT-based generative model with multi-objective optimization for de novo drug design. Unlike traditional virtual screening that searches existing libraries, our generative approach explores uncharted chemical space to create entirely novel, synthesizable molecules.

The framework integrates machine learning-based binding affinity predictions with periodic docking scores, guiding the RL agent towards chemically diverse, target-specific molecules. Retrosynthetic pathway predictions confirm the synthetic accessibility of generated compounds, accelerated by our Enalos Asclepios KNIME nodes.

Published: "Hybrid RL-driven de novo design of BRAF inhibitors with GPT model and retrosynthetic feasibility"
GPT-Based Generation Pretrained transformer model generates valid, drug-like SMILES strings with high chemical diversity
Hybrid Reward Strategy Combines ML affinity predictions with docking scores for efficient target-specific optimization
Scaffold Novelty Generates molecules with enhanced scaffold novelty compared to affinity-only reward strategies
Retrosynthetic Feasibility Automated pathway predictions confirm synthetic accessibility of top candidates
BRAF Validation Top-docked compounds show pharmacophore-level similarity to known BRAF kinase inhibitors

See It in Action

Real-world projects where our pipeline delivered validated results

Case Study

RNSMOKE: Natural CYP2A6 Inhibitors for Smoking Cessation

Virtual screening of 700,000+ natural products, ML modeling with Isalos, and in vivo validation identified novel compounds with IC50 of 0.64 µM — comparable to the gold standard.

Read Case Study
Case Study

ChemBioAD: Multi-Target BACE Inhibitors for Alzheimer's Disease

Structure-based screening, consensus ML models, and multi-target validation discovered a potent BACE1 inhibitor (IC50 = 160 nM) with dual activity against tau aggregation.

Read Case Study
Case Study

ChemBioCOMT: Novel COMT Inhibitors for Parkinson's Disease

Ligand-based ML, structure-based docking of 14,400 molecules, and in vitro/in vivo testing discovered a novel COMT inhibitor (IC50 = 8.1 μM mCOMT) with zero cytotoxicity.

Read Case Study
Case Study • Featured in Media

Small-Molecule PPI Dual Inhibitors of TNF & RANKL

Cheminformatics pipeline screened ~15,000 molecules to discover T8 & T23 — dual TNF/RANKL inhibitors with low toxicity. Featured in EurekAlert & MS News Today.

Read Case Study
Case Study

Plant-Origin Natural Product Inhibitors of TNF & RANKL

In silico screening of plant-origin NPs discovered the first natural product TNF inhibitors, including A11 (Ampelopsin H) — the first NP dual TNF/RANKL inhibitor.

Read Case Study
Case Study

Enalos Asclepios Pipeline for TNF Inhibitor Discovery

Advanced Enalos Asclepios KNIME nodes with 1000ns MD simulations identified Nepalensinol B, which completely abolished TNF-TNFR1 binding.

Read Case Study

Ready to Accelerate Your Drug Discovery?

Let's discuss how our computational pipeline can help you find better candidates faster.

Get in Touch