Applied AI in Bioinformatics for Mutational Signature Discovery in Viruses and Cancer
School of Engineering and Technology
Dr Ayub Bokani
Synopsis
Every process that damages DNA, whether ageing, UV light, smoking, or the activity of enzymes such as APOBEC, leaves behind a recognisable pattern of mutations. Reading these patterns tells us what caused a disease and can guide how it is treated. This project uses machine learning and bioinformatics to discover, validate and interpret mutation signatures across biological data, and to link them to donor and clinical information so the results can support precision medicine, infectious-disease surveillance and public health. The scope is broad and can be shaped around a student's interests, spanning cancer genomics, viral genomes, new detection methods, and the pipelines that make signature analysis reliable and reproducible. One example already underway is MutSig-Cloud, an open-source, browser-based pipeline built within CQUniversity's CML-NET Biomedical Cluster with partners in the USA and Japan, but students may equally focus on the biology of a particular disease, on developing new methods, or on benchmarking and validation. Whatever the direction, students join an active international collaboration with work in Nature Communications and a current Nature Genetics submission.
Biological Sciences; Information and Computing Sciences; Medical and Health Sciences
bioinformatics, machine learning, mutational signatures, cancer genomics, viral genomics, APOBEC, precision medicine, computational biology
Available now (flexible)
Doctorate; Either Masters or Doctorate; Masters
Sponsor
Texas Biomedical Research Institute (USA), Kumamoto University (Japan)
Other Special Notes
Possible directions include: discovering and cataloguing mutation signatures in cancer genomes; identifying signatures in viral genomes such as HIV-1 and SARS-CoV-2, where no reference catalogue yet exists; comparing viruses and cancers to uncover shared mechanisms such as APOBEC editing; developing and benchmarking new machine-learning methods for signature detection and validation; linking signatures to demographic and clinical data for precision medicine; and building open, reproducible pipelines and tools, for example the MutSig-Cloud platform. A related computational-health direction is decoding neural signals for brain-computer interfaces. A working prototype and international collaborations are already in place through the CML-NET Biomedical Cluster.
