Very cool work from Yang Lu et al. demonstrating miscalibration of BLASTP’s E-values and generating well-calibrated values via a knockoff-based approach (cc @mikelove.bsky.social) - https://academic.oup.com/bioinformatics/advance-article/doi/10.1093/bioinformatics/btae729/7916501! More analyses could benefit from knockoff-based approaches.
Comments
Can you clarify what you mean by knock-off based approaches? Is it error rate calibration via sampling the null distribution?