This drug discovery startup hits a new high in AI docking.

London-based RECEPTOR.AI recently unveiled ArtiDock v1.5, their latest advancement in AI docking software. The update expands ArtiDock's capabilities, especially in immuno-oncology.

The Receptor.AI’s computational platform was successfully validated in several projects on protein-protein interaction (PPI) targets, traditionally addressed only by antibody-based modalities.

Notably, the release of ArtiDock v1.5 aligns with the introduction of the PoseBusters v3 dataset, which challenges and evaluates AI docking technologies. The latest version of PoseBusters has become even harder for AI models.

When presented with the PoseBusters v3 challenge, ArtiDock v1.5 stands out, showing improvements over its previous version and competing technologies. The AI ArtiDock v1.5 excels when tested against the PoseBusters dataset's diverse structural metrics.

These studies highlight ArtiDock’s effectiveness across various levels of precision, as indicated by different RMSD cutoffs.

ArtiDock v1.5's inference speed is a standout feature, matching other leading deep learning models while delivering accurate results efficiently. For instance, ArtiDock 1.5 is one order of magnitude faster than Glide, Alpha Fold-Latest and two orders of magnitude faster than Vina and Gold, which makes it preferable for high-throughput screening applications. In general, ArtiDock outperforms all existing techniques in terms of throughput and pose quality.

Receptor.AI informed me that NVIDIA BioNeMo, a leading cloud service in drug discovery, is set to include ArtiDock v1.5 in its suite of tools. This integration would position ArtiDock v1.5 alongside notable offerings from DeepMind and MIT, marking it as one of the first commercial AI models to be featured on the BioNeMo platform (link to the NVIDIA presentation in the comments).

Check out a technical case demonstrating ArtiDock’s performance benchmark vs leading docking technologies (link in the comments).

#shorts #science #drugdiscovery #biopharmatrend #WhereTechMeetsBio #artificialintelligence #docking

[WhereTechMeetsBio]

London-based RECEPTOR.AI recently unveiled ArtiDock v1.5, their latest advancement in AI docking software.

The update expands ArtiDock's capabilities, especially in immuno-oncology.



The Receptor.AI’s computational platform was successfully validated in several projects on protein-protein interaction (PPI) targets, traditionally addressed only by antibody-based modalities.



Notably, the release of ArtiDock v1.5 aligns with the introduction of the PoseBusters v3 dataset, which challenges and evaluates AI docking technologies.



The latest version of PoseBusters has become even harder for AI models.



When presented with the PoseBusters v3 challenge, ArtiDock v1.5 stands out, showing improvements over its previous version and competing technologies.

The AI ArtiDock v1.5 excels when tested against the PoseBusters dataset's diverse structural metrics.

These studies highlight ArtiDock’s effectiveness across various levels of precision, as indicated by different RMSD cutoffs.



ArtiDock v1.5's inference speed is a standout feature, matching other leading deep learning models while delivering accurate results efficiently.

For instance, ArtiDock 1.5 is one order of magnitude faster than Glide, Alpha Fold-Latest and two orders of magnitude faster than Vina and Gold, which makes it preferable for high-throughput screening applications. In general, ArtiDock outperforms all existing techniques in terms of throughput and pose quality.



Receptor.AI informed me that NVIDIA BioNeMo, a leading cloud service in drug discovery, is set to include ArtiDock v1.5 in its suite of tools.

This integration would position ArtiDock v1.5 alongside notable offerings from DeepMind and MIT, marking it as one of the first commercial AI models to be featured on the BioNeMo platform (link to the NVIDIA presentation in the comments).



Check technical case demonstrating ArtiDock’s performance benchmark vs leading docking technologies (link in the comments):