BOL

Imagine comparing humans and chimpanzees to discover which genes make us different. You have millions of RNA-sequencing reads—but there is a catch: the computer may align these reads more easily to one species' genome than the other.

Suddenly, a technical error can look like a biological difference.

This is the problem Kenneth A. Barr and Yoav Gilad tackle with CrossFilt, a new computational tool introduced in Genome Biology journal.

CrossFilt acts like a careful referee. It keeps only sequencing reads that can be reliably compared across species, filtering out reads that could create alignment bias.

When tested on human, chimpanzee, and rhesus macaque data, CrossFilt reduced false discoveries and produced more reliable cross-species comparisons.

The message is simple: before asking what makes species different, we need to make sure our methods aren't creating those differences.

In genomics, sometimes the most important discovery is finding—and removing—the bias hiding in the data.

Read more at https://link.springer.com/article/10.1186/s13059-026-04082-2?