Launch of Malaria Inhibitor Prediction platform (
Machine learning models for predicting compounds that inhibit blood stage malaria are available as a web service (https://www.ebi.ac.uk/chembl/maip). James Duffy (MMV) and Nicolas Bosc (EMBL/EBI) introduce the malaria inhibitor prediction platform (MAIP) as a free public service, open to all scientists. MAIP is the result of a public-private collaboration to develop a consensus model trained with data from five pharma and not-for-profit partners on their private datasets. Models were combined by EMBL-EBI and the service can be used to identify compounds for screening.
3.00 min: James describes the background to the project and highlighted The Medicines for Malaria Venture’s commitment to supporting Open Innovation and taking malaria drug discovery to the next level. MMV are offering to test a sample of any compounds that you identify using MAIP in their Plasmodium falciparum asexual blood stage assay. Please contact them [email protected] for further details – terms and conditions are available on the MMV website.
15.00 min: Nicolas focused on the development of the machine learning consensus approach which combines models generated from large compound collections to increase resulting model predictivity. This approach allowed models to be shared without revealing the composition of the 11 training sets, a total of 6 million compounds. He went on to give a demonstration of running MAIP online. The service is primarily designed for hit identification by virtual screening and can handle up to 1 million compounds in a single run.
39.00 min: Mark Gardner (AMG Consultants) described initial screening results obtained from testing two sets of compounds, comparing one selected using MAIP against a random selection.
41.00 min: Q & A session
