Novel semi-supervised Bayesian learning for impurity detection in oligonucleotide drugs

点击翻译按钮获取中文摘要

ObjectivesDevelop Bayesian data science methodology to profile oligonucleotide impurities automatically, accurately, and delivers statistical measures of certainty.DetailsOligonucleotides are large complex molecules they are currently very difficult to profile for impurities, as the analysis is labour intensive and the data complexity is high. This project will aim to apply data science methodology to carry out impurity profiling over a larger number of batches of mass spectrometry data that will enable us to learn how to characterise the known oligonucleotide signal and deconvolute it from a number of known and unknown impurities longitudinally, in a semi-supervised learning framework, to gain detailed insights towards these impurities longitudinally; while confirming the overall consistency of the profile, identify any change patterns, trends over batches, and any correlation between impurities. This approach can also provide semi-quantitative result to aid our understanding in key studies such ASAP, terminal sterilisation and forced degradation for both drug substance and drug product. DeliverablesBy establishing a data analytics pipeline and embedding it as part of our routine analysis, monitor impurities more closely and more precisely. Apply the knowledge to identify possible issue in manufacturing and improve process chemistry by pinpointing impurities associated with different steps of the synthesis.This project falls within the EPSRC statistics and applied probability research area, and is a collaboration with AstraZeneca.