Lin Xu

BOSE estimates sample mean and standard deviation from median, range, and interquartile range using Bayesian order statistics, showing robustness and improved performance in small samples and non-normal distributions.

The paper identifies dynamic regulatory network hubs in human lineage specification using Perturb-Seq and constructs a deep learning model to predict perturbed TFs in patient transcriptomes.

This study benchmarks and optimizes Perturb-seq in human pluripotent stem cells, identifying shared regulatory networks during cardiomyocyte differentiation.

This study performs GWAS on BMD traits and fracture risk using 30,000 UK Biobank samples, identifies novel loci, and prioritizes drug targets for osteoporosis prevention.