The area under the precision-recall curve (AUPRC) is a valuable metric for quantifying classification performance, particularly in situations with imbalanced classes like cancer diagnosis and cell type annotation. A research team, led by Prof. Stephen Kwok Wing Tsui and Prof. Qin Cao from the School of Biomedical Sciences (SBS) at CUHK, along with Prof. Kevin Yip from the Sanford Burnham Prebys Medical Discovery Institute in the USA, has discovered that the AUPRC values computed by 10 commonly used software tools, collectively used in more than 3000 published studies, rank classifiers differently and some tools yield overly-optimistic results. This suggests the potential invalidity of some published biological findings. This research has been published in Genome Biology: https://genomebiology.biomedcentral.com/articles/10.1186/s13059-024-03266-y.

The study was conducted by Ms. Wenyu Chen, a first-year PhD student at SBS, and Mr. Chen Miao, a Research Assistant at SBS. Both Wenyu and Chen are graduates of the MSc program in Genomics and Bioinformatics at SBS.