Warning!
Your browser does not support HTML5.
Display of the content may be incorrect. Please make sure that your browser supports HTML5 syntax.
Search for Small Molecules by SMILES, name, CAS, MFCD, ID, InChI Key
Search for Biologics by ID, Target name, CAS, Protein name, Gene ID, Clone ID, Source Organism
0
0.00 €
Currency
Home Discovery CRO Machine Learning based Services

Machine Learning-based Services

Our machine learning-based services are designed to revolutionize the way you explore, analyze, and uncover insights in your data. Whether it's predicting potentially active compounds, exploring vast datasets, or looking for starting points for your projects, our services offer practical solutions that put you at the forefront of drug discovery innovation.

ML-driven hit expansion

ML-driven hit expansion image
ML-driven hit expansion is becoming integral to the modern Drug Discovery process, enabling researchers to efficiently explore a vast chemical space and identify promising compounds for further investigation. By leveraging machine learning algorithms, you can efficiently analyze large datasets after HTS/DEL screening, pick true positive compounds, and based on learned patterns expand selected hits by compounds from commercial spaces like Chemspace In-Stock screening collection, Freedom Space, or Enamine REAL Space.
ML-driven hit expansion image

ML-supported data analysis

ML-supported data analysis image
ML-supported data analysis in the context of Drug Discovery involves the application of machine learning techniques to analyze complex biological, chemical, and pharmacological data. This approach leverages various ML algorithms to reduce noise, uncover patterns, relationships, and insights from diverse datasets, including results from biological assays.
ML-supported data analysis image

DiffDock

DiffDock image
DiffDock is considered the state-of-the-art method for blind docking. It is a generative diffusion model which is specialized in protein-ligand interaction prediction. It implements an excellent example of “inductive bias” - generating molecular conformers by focusing on torsion angles but keeping bond lengths and angles constant. [1]
DiffDock image

Physicochemical properties prediction (logP, logD, solubility(water/DMSO))

Physicochemical properties prediction image
The physicochemical properties of compounds are the key factors in the drug design process. It is important to have a valid tool to predict structure-property relationships to understand and model the action of selected compounds in your projects.
Physicochemical properties prediction image

Active learning based on the property of interest

Active learning based on the property of interest image
We offer to utilize active learning strategies, including various sampling techniques and model selection methods, to iteratively select and prioritize the most informative compounds for experimental testing or further analysis. It is used to speed up and reduce the costs of computationally challenging approaches like docking or molecular dynamics. Such a combination allows performing expensive calculations only on a fraction of the available space (about 1%), while still retaining most of the top-performing compounds.
Active learning based on the property of interest image

Custom project requests

You have a project idea in mind that is not covered on this page? We'd love to hear it! Our expert machine learning team will be happy to discuss any of your ideas and enhance your journey towards groundbreaking pharmaceutical solutions. Fill in the form below and we'll contact you within 24 hours to schedule a discovery call.

Would you like to get more details?

Fill out the form to contact our team
Please fill out this field
Please enter a valid email address
Please fill out this field
Please fill out this field
Please fill out this field
Thank you, we will contact you shortly
An error occured
References
  1. Corso, G.; Stärk, H.; Jing, B.; Barzilay, R.; Jaakkola, T. DiffDock: Diffusion Steps, Twists, and Turns for Molecular Docking. 2022. https://doi.org/10.48550/ARXIV.2210.01776