A Survey on Cancer Subtyping Based on Deep Learning Using Pan-Cancer and Multiomic Data

Abstract

Tumor subclasses with clinical implications are identified by breast cancer gene expression patterns. In this study, ROBERT TIBSHIRANI et al. make a suggestion. Based on changes in gene expression, the tumours were divided into three groups: basal epithelial-like, ERBB2-overexpressing, and normal breast-like. Two separate gene sets, one representing a collection of 456 cDNA clones originally chosen to reflect intrinsic tumour features and the other being a gene set that was highly correlated with patient prognosis, were clustered to reveal that both groupings were quite robust. The basal-like subtype had a dismal prognosis, and the two oestrogen receptor-positive groups had significantly different outcomes, according to survival studies on a sub cohort of patients. In this study, three fibroadenomas and 78 breast carcinomas were examined. This collection comprises of 40 tumours that have already been studied and characterised. A total of 85 tissue samples from 84 individuals were examined.