Atomwise (AIMS Platform)

Pioneer of deep learning for virtual drug screening, founded 2012. Atomwise’s AIMS (Atomnet Interactive Modeling System) platform uses deep convolutional neural networks on 3D molecular structures to predict binding affinity between compounds and protein targets. Has screened 10B+ compound-target pairs across 750+ drug discovery programs.

Key Capabilities

• AtomNet: Deep CNN trained on millions of bioactive compounds from BindingDB, ChEMBL

• Billion-compound virtual screening: Screens Enamine REAL and proprietary libraries

• AIMS platform: Iterative hit-to-lead optimization with AI-guided design cycles

• Academic partnership program: AIMS for Academics — free access for non-commercial research

Deep Learning Virtual Screening for Neurodegeneration Targets

A landmark study demonstrated that AI-based virtual screening is a viable alternative to high-throughput screening across 318 diverse targets, achieving hit rates comparable to experimental approaches at a fraction of the cost and time

. This finding has significant implications for neurodegeneration drug discovery, where targets such as tau, alpha-synuclein, and TDP-43 have proven difficult to screen using conventional methods due to the challenges of working with disordered or aggregation-prone proteins.

Atomwise-style deep learning approaches have been applied to identify small molecule inhibitors across a range of therapeutic targets, including targets relevant to neurodegeneration such as protein kinases involved in tau phosphorylation and enzymes implicated in neuroinflammatory cascades

. The ability to screen billions of virtual compounds enables researchers to identify novel chemotypes that would be missed by traditional screening libraries, which are often biased toward well-characterized chemical space. Inhibition of casein kinase 1 delta has recently been identified as a novel therapeutic strategy for amyotrophic lateral sclerosis through computational screening approaches, illustrating how virtual screening can reveal previously unrecognized therapeutic opportunities in neurodegeneration .

The AIMS for Academics program has been particularly impactful for neurodegeneration research, providing academic laboratories with access to industrial-scale virtual screening capabilities that would otherwise be prohibitively expensive. This democratisation of AI-driven drug discovery has enabled smaller research groups to pursue target-based screening campaigns against neurodegeneration-related proteins, contributing to the identification of novel chemical probes and lead compounds.

Track Record

• 750+ drug discovery programs across academia and pharma

• Neurodegeneration focus: programs in ALS, multiple sclerosis, neurological targets

• Several compounds advanced to preclinical and clinical stages via Atomwise-partnered programs

Relevance to SciDEX (Neurodegeneration)

Atomwise is a direct reference case for AI-driven neurodegeneration drug discovery. Their AIMS for Academics program could enable SciDEX community members to screen compounds against neurodegeneration targets identified through SciDEX debates. The AtomNet architecture provides a benchmark for evaluating Forge tool implementations.

Cross-References

• [[Virtual Screening]]

• [[Deep Learning Drug Design]]

• [[Diseases: Neurodegeneration]]

• [[ALS Therapeutics]]

• [[Drug Discovery AI]]