Drug Repurposing

Drug Repurposing

 Drug repurposing is a process of locating other indications of already approved, shelved, or failed drugs. De novo drug development requires approximately 12~15 years with $2.5~$3 billions of investment. Yet the chances of success of de novo drug development are less than 10%.

Drug repurposing methods reduce the development time to 3~5 years and investment to few hundred millions.

We at BioSpero decentralize drug development technology with state of art drug repurposing techniques built on novel and robust data-driven methods. Our aim is to eliminate failures and waste when translating drug candidates to clinical trials using AI-driven, multi-disciplinary quantitative methods.

Drug Repurposing Workflow

BioSpero’s SperoPredictor is unique drug repurposing framework which integrates multiple machine learning models, Trained on various aspects of drug and diseases. Fast, precise, and reliable drug repurposing platform is rigorously trained And validated in multiple practical applications. Predictions through ML enabled repurposing framework are passed through multiscale molecular docking prior to enter in preclinical testing. Thus, SperoPredictor presents a reliable and accurate strategy with promising results.


Details of Repurposing Workflow

Different type of public and in-house biodata related to drugs and diseases is collected. Based on the data curation, integration, and design knowledge BioSpero experts integrate and analyze data to make it machine learning (ML) ready. The curated data additionally is used in other fields like discovering biomarkers.

BioSpero has AI technology (SperoPredictor) that enables drug repurposing of FDA approved drugs based on the training and validation on databases such as drugs, genes, diseases, and other clinical information. The state of art algorithms are fast, accurate, and reliable with high positive rate maximizing the chances of success.

BioSpero’s multilayered virtual screening and molecular docking technology further adds in the effectivities of AI based repurposing framework. 

The repurposable drugs molecules predicted by BioSpero’s SperoPredictor are then passed through high throughput and multiscale virtual screening framework to further prioritize the key potential drug compounds for the indication and vice versa.

Specifications & Applications

Specifications

• SperoPredictor is a machine learning (multiple models) and multiple docking protocols-based ensemble Drug repurposing platform

• Machine learning part of the SperoPredictor is rigorously trained on the diverse aspect of the drugs and disease data.

• The training and testing data is consisted of:

  a. Four frug features for 2875 FDA approve Drugs

  b. Three disease features for 2400 Disease indications

•  The data is integrated from multiple databases.

•  SperoPredictor can find the alternative indications of the FDA approved off-patent drugs with high prediction confidence.

•  The Predictions of SperoPredictor are validated through multi-layered validation pipeline. The validation steps are:

• Literature based validation

• Multiple docking protocols-based validation


Applications

• SperoPredictor can be deployed to find the repurposed drugs for any given disease of interest.

• SperoPredictor also has the ability to predict the new indication for a given small molecule drug.

• Additionally, the SperoPredictor has been deployed in multiple use cases. Such as:

• Fibrosis – 10 Drugs were predicted out of which 3 drugs (33%) were verified from the literature.

• NASH – 27 Drugs were predicted out of which 7 (25%) were verified from literature.

• Covid-19 - 25 Drugs were predicted out of which 12 Drugs (48%) were verified from Literature. The predicted drugs contained the FDA Approved Remdesivir, and 2 drugs with status of Emergency use Authorization.