BCM and Rice publish new computational tool for diarrheal disease classification in top clinical journal

Taxa4Meta represents a new “best practices” approach for more accurate diagnoses

BCM & Rice Researchers

After seven years of research, Baylor College of Medicine researchers Qinglong Wu, Shyam Badu, Sik Yu So, Dr. Tor C. Savidge along with Professor Todd. J Treangen from Rice’s Department of Computer Science introduced a new method, Taxa4Meta, to identify gut bacteria associated with diarrheal disease. Their new findings were published in The Journal of Clinical Investigation.

The focus of Taxa4Meta is to provide accurate classifications of gut bacteria, whether they are disease associated or not. Taxa4Meta is able to pull from decades of data to provide more accurate sequences for diagnosing patients with gut disease. 

“The field has been involved in shortening sequence lengths to provide abundance analysis. The problem with that is it tends to contribute to misdiagnosis. The one difference we are using [in Taxa4Meta] is being able to exploit all sequences that are there with the goal of providing a more accurate diagnosis,” said Dr. Savidge, professor of pathology and immunology at Baylor. “[When you have susceptibility], if you understand which bacteria are there, or missing more importantly, you can start to understand the mechanisms that are involved in the disease process.” 

With current microbiome classification methods, however, the limitations prevent individual research projects from being applied on a larger scale.

“The field has largely lacked known-truth benchmarking datasets,” Professor Treangen said. “As a field, we often publish tools based on performance specific to simulated data sets, which are important for establishing baseline performance. The problem with depending on simulated datasets alone is that they are an approximation of datasets from clinical samples; often when we compare and contrast taxonomic profile performance on simulated data and real samples, there's a substantial disconnect with respect to performance.”

“For example, within the Texas Medical Center, you can develop measures that are reasonable within individual hospitals,” Dr. Savidge said. “These tend to fall apart a little bit when you move from hospital to hospital in a different demographic.”

One key bacteria that Taxa4Meta identifies is C. difficile, a bacterium found in the colon, often linked to diseases after a patient takes antibiotics.

“[C. diff] is listed as one of the CDC Center of Disease Prevention’s top priority pathogens [to identify directly]. This is one of the diseases where we know that the microbiome plays a very important role,” Dr. Savidge said. “Microbial therapy is an emerging field. To accurately classify which bacteria are missing in patients that are susceptible to C. diff means that we can put those bacterias back in as a treatment.”

Both professors mention their gratitude to former Baylor instructor and lead author Qinglong Wu for his persistence over seven years.

“While I was thrilled to be involved on this clinically relevant project to provide computational advice and perspective, it's really Dr. Wu's blood, sweat, and tears over seven years that made this project successful,” Professor Treangen said.

The success of Taxa4Meta showcases the strength of Rice University collaborations with neighboring Houston medical centers.

“One of my main goals when arriving to Rice was to strengthen and further collaborative efforts between Rice and the TMC; science in its best form is a team game.” Professor Treangen said.

This research was funded by the National Institute of Allergy and Infectious Diseases. For additional details on the team’s research, please visit the NIH website




Ariana Wang, Contributing Author