Find the best drug candidates for your cancer targets || Deep Origin

Proprietary models for better drug discovery

Simulate living systems

Predict ligand properties (synthesizability, solubility, etc.)

Target active or allosteric sites

Structure-free docking for protein structure prediction

Broader ligand property prediction including ADMET properties

Multi-level (and scale) models

Easily scale and port compute, analysis, and collaboration

Docking performance across a spectrum of target classes

Benchmarked on the PDBBind 285 set (1)

ref. 1. https://pubs.acs.org/doi/10.1021/acs.jcim.8b00545, benchmark data available at http://www.pdbbind.org.cn/index.php, used for comparability with prior results

Cheminformatics performance benchmarks

logP

Rank

Model

RMSE

OCHEM

0.34

0.49

Biosim Props

0.449

DNN (taut)

0.47

DNN (mono)

0.5

ACD / GALAS

0.5

ALOGPS

0.5

KOWWIN

0.65

JChem

0.72

logD

Rank

Model

MAE

Biosim Props

0.425

Chemprop - RDKit

0.466

Chemprop

0.469

BaseBoosting

0.479

ContextPred

0.535

GCN

0.541

AttrMasking

0.547

NeuralFP

0.563

AttentiveFP

0.572

logS

Rank

Model

MAE

Biosim Props

0.525

Chemprop-RDKit

0.762

AttentiveFP

0.776

Chemprop

0.818

RDKit2D + MLP (DeepPurpose)

0.827

Basic ML

0.828

GCN

0.907

NeuralFP

0.947

CNN (DeepPurpose)

1.023

ref. 1. TDC.Solubility_AqSolDB
2. Ulrich N., Goss K. U., Ebert A. Exploring the octanol–water partition coefficient dataset using deep learning techniques and data augmentation // Communications Chemistry. – 2021. – Т. 4. – №. 1. – С. 90.
3. TDC.Lipophilicity_AstraZeneca

You've got the target, we'll find you the drug candidate

Utilizing our innovative virtual screening technology, we can accelerate your drug discovery program. Our experts can partner with you through each step of the process.

Small molecule discovery

Target Structural Dynamics

No crystal structure? Need to model or distrupt protein-to-protein interaction? We've got you.

Hit Identification

Find the best binders, account for multiple physicochemical and ADMET properties.

Counterscreen against antitargets

Screen for binding against undesired targets to narrow hits and avert safety issues.

Lead Optimization

Optimize lead compounds across multiple parameters.

Biologic and large molecule discovery

Target structural dynamic predictions

Understand the fluctuation in your target's structure over time.

De novo antibody design

Design antibody variable regions to fit the antigen.

Antibody-antigen binding prediction

Predict binding conformations of existing antibodies against an antigen.

Find the best drug candidates for your cancer targets

Proprietary models for better drug discovery

Simulate living systems

Docking performance across a spectrum of target classes

Cheminformatics performance benchmarks

You've got the target, we'll find you the drug candidate

Small molecule discovery

Biologic and large molecule discovery

Request access

We value your privacy