Predicting resistance to antituberculars in global cloud-based platform

Описание: Philip Fowler from University of Oxford:
The SARS-CoV-2 pandemic has accelerated the transition of clinical microbiology and public health to big-data, genetics disciplines. We have successfully created, together with Oracle, the Global Pathogen Analysis Service (GPAS, https://gpas.cloud) which allows clinical laboratories to upload the raw files output by genetic sequencers which are then processed at scale in the cloud using Kubernetes, returning the SARS-CoV-2 lineage and consensus genome in less than fifteen minutes. GPAS is free to low- and middle-income countries. Before the end of 2022, we will have added the capability to upload and process tuberculosis samples, returning epidemiological information and an antibiogram. The latter is a list of which antituberculars the infection will be susceptible to and to which it will be resistant. Since this is derived from a dataset of existing samples, it cannot always return a definite prediction and hence we have also trained machine learning models that use genetic, structural, chemical and evolutionary features to predict the effect of rare and novel mutations on the action of a specific antibiotic. The goal is to deploy our growing suite of models within the tuberculosis pipeline in GPAS, ensuring clinicians have sufficient information to make an informed decision about initial treatment.