BI assistant scientist develops a tool for stratifying cancer patients
Bioinformatics assistant scientist Tongjun Gu Ph.D. has developed an advanced deep learning method to stratify cancer patients into clinically relevant subgroups and further identify the potential biomarkers between the subgroups. This will help with the development of personalized treatments and prognosis.
The method can integrate multi-platform datasets like gene expression, miRNA expression, methylation, copy number alteration, and protein expression. It can also do transfer learning between different cancers. For example, in cancer types with smaller sample sizes, information can be borrowed from a cancer type with a larger sample size to subgroup it.
The advanced method was published in Nature Research’s Scientific Reports journal in an article titled “Integrating multi-platform genomic datasets for kidney renal clear cell carcinoma subtyping using stacked denoising autoencoders.” The associated tool is open-access and can be accessed on GitHub.