Every drug deserves a second purpose.

A motif-and-pocket-aware screening engine that predicts drug-target binding affinity from molecular structure, protein geometry and sequence — backed by PerceiverIO and multi-pocket scanning.

What happens when the discovery phase shrinks from years to weeks.

Drug development is built on a 13-year pipeline and a 90% failure rate. Repurposing approved compounds skips the longest, most expensive years — but only if the screening is fast and the predictions are explainable.

$0B
Cost per new drug
What every approved compound currently costs to bring to market
0 yrs
Traditional pipeline
Discovery → trials → confirmed effectiveness → approval
0 yrs
With computational repurposing
Discovery phase compressed from 6 years to 1
0+
Approved drugs
Already safety-tested, already manufactured, largely unscreened
ECONOMIC

90% lower cost

Repurposing skips synthesis, toxicity profiling, and Phase I safety data — the most expensive years of new-drug development. The same indication can be validated for a fraction of the original investment.

CLINICAL

Patient access in years, not decades

Approved compounds carry known safety profiles. A confirmed repurposing hit can move directly into Phase II trials, cutting roughly five years off the path to a new indication.

SCIENTIFIC

Coverage of under-explored disease space

Rare diseases, neglected tropical diseases and pandemic targets rarely justify a new-drug program. Repurposing makes them economically viable — and DrugRepAI makes the screening tractable.

Three layers. One verifiable answer.

Most tools collapse drug-disease scoring into a black box. We split it into three named layers, each with a separate output you can audit.

01 / 03
01

MotifGT-DTI

Encodes drugs as pharmacophoric motifs (BRICS) and proteins as pocket-aware structures. Cross-attention produces a binding score and motif–residue map.

Motif-level functional units
Physical binding score 𝑠(𝑑, 𝑡)
Attention map A — chemical XAI
No collapse: every coord distinct
Transfers across scaffolds
02

Disease in protein space

Disease profiles are encoded as sparse protein vectors from gene–disease evidence. Drug and disease representations become directly comparable.

Drug Φ(𝑑) in protein space
Disease Ψ(𝑚) in protein space
Bridge = inner product
Per-protein attribution built in
03

Binding-Anchored Bridge

The bridge combines binding signals, disease associations, and protein topology into a unified score. A lightweight learned head refines the final prediction.

Drug profile Φ(d)
Disease profile Ψ(m)
PrimeKG topology
Binding-aware scoring
Multi-hop propagation

Drug-Target Screening

Simulate AI-powered molecular screening and explore binding affinity predictions in seconds.

Input / Molecular data
Molecular structure
Aspirin example
Protein 3D geometry
3HKI.PDB
Amino acid sequence
Hemoglobin example
Output / Binding affinity ● 1.42S · OK
— Score → float ∈ [0.0, 1.0]
0.7296
MODERATE AFFINITY · worth investigating
0.0 WEAK STRONG 1.0
Pipeline info STACK V0.4.2
Model PerceiverIO + Multi-Pocket GNN
Drug encoding BRICS Motif Graph (DGL)
Protein encoding41-dim residue + Cα-Cα
Sequence One-hot [1, 1200, 26]
Output Float ∈ [0.0, 1.0]

Different tools solve different parts of the problem.

DrugRep AI combines repurposing focus, modern graph architecture, and multi-pocket scanning — the only stack built end-to-end for in-silico repurposing.

Tool Approach Focus Multi-pocket Note
DeepDTA Sequence DTA Binding affinity No Predates graph-based drug models
Atomwise CNN docking Hit discovery Limited Focused on de-novo, not repurposing
Insilico Medicine Generative Novel chemistry No De-novo design, not existing drugs
Recursion Phenotypic screen Wet-lab at scale No In-vitro, not computational
REPO4EU Knowledge graph Repurposing No Public consortium, limited AI depth
DrugRepAI GNN + PerceiverIO Repurposing Yes Graph + multi-pocket + transformer fusion Knowledge Engine, not just Prediction
0
Approved drugs
in DrugBank library
0 faster
Discovery time
vs de novo development
0%
Cost reduction
vs $2.2B per new drug
0%
Trial failure rate
what we're trying to fix

What researchers, biotech teams and consortia are saying.

We built DrugRepAI for the people who need to know why, not just what. Here's how that's landing in the labs and pipelines using it.

We don't do this alone.

DrugRepAI sits inside an ecosystem of scientific consortia, public data sources and infrastructure partners. Each one is named because each one is verifiable — same principle as our predictions.

Moods Dev Studio
Provide end-to-end product development — from ML architecture, engineering and product design
Disgenet
Provide AI-powered genomics intelligence for biomedical research and drug discovery.

Driven by Talent.
United by Innovation.

Our team brings together expertise in design, technology, and product strategy to create modern solutions for forward-thinking businesses.

Explore the team

Book a 30-minute call.

Whether you're testing a hypothesis, expanding a discovery pipeline, or scoping a consortium use case — start with a conversation. No deck, no demo theater, just your problem and what we can do about it.