MegaTrans - human transporter machine learning models

Summary Being able to predict interactions with important human transporters would be of value to new drug design to avoid compounds that interact with them and cause undesirable side effectsOATP BSLCO Band OATP BSLCO Bare `uptakeandapostransporters largely restricted to the sinusoidal aspect of hepatocytesThey both transport a wide variety of structurally unrelated compoundsincluding members of several clinically important drug families such as statinssartans and angiotensin converting enzymeACEinhibitorsWe now propose to test overdrugs againstsubstrates for each transporter in vitroWe will then use these data to curate and validate machine learning modelsWe will also use an array of machine learning methods as well as multiple model evaluation metricsThis will enable us to develop a web based software tool called MegaTrans that will encourage the user to input their own compound structures and generate predictions for interactions with transporter s of interest and then visualize the similarity to the training set of each model using several different visualization methodsThe return on investment of such a tool would be that it could assist in the design and selection of more favorable compounds that avoid transporters of interest while also saving time and moneyIt could also identify compounds that are already approved that might present a drug interaction riskPredicting such behavior seen in vivo is ideal and will lead to the prioritization of compounds to test in vitro for potential drug drug interactionsIn Phase II we would greatly expand the number of transporters which we would generate data on and build models such that we could address all the major transporters of interest to drug discovery Narrative The objective ofMegaTransis to develop a new computational system and tools for integrating human transporter data into drug discovery pipelinesas well as enabling its analysis and visualizationThis will then enable improved computational tool development for in vitro to in vivo extrapolation of xenobiotic exposures across a range of assay typesIt will also assist with developing computational tools for quantitatively modeling drug drug interactions of xenobioticsAcross Phase I and Phase II we will generate data for transporters which currently have a paucity of data in the public domainuse validated machine learning algorithmsaccess the latest curated datasets and develop a user intuitive interface and visualization system to enable predictions