COSAM News Articles 2023 May Auburn statistician works to help identify gene expressions to predict breast cancer

Auburn statistician works to help identify gene expressions to predict breast cancer

Published: 05/09/2023

Rob Molinari, an assistant professor in the Department of Mathematics and Statistics, is working on a new idea in data analysis that could directly help identify genetic expressions linked to breast cancer.

He is part of an international team seeking to use data to help accurately use statistical models to determine if a person has genetic data predisposed to breast cancer.

“We are interpreting the same reality through multiple statistical models, just like the same emotion (for example) can be expressed through different sentences,” Molinari said. “Think of these two sentences: Today is gorgeous; and Today is beautiful. Both basically have the same meaning, however, data analysis on the contrary admits only one true final model (sentence) instead of allowing multiple interpretations of the same reality.”

Molinari is working on an algorithm to change that. Then, each model (sentence) could have different expressions but deliver the same interpretation (meaning), which could unlock essential findings in different fields such as genomics for cancer detection and treatment.

“When the algorithm is applied to breast cancer data, it is looking for which combinations of genetic expressions are linked to the presence of breast cancer and how it uses miRNA data — expressions underlying those behind the Covid vaccine,” Molinari said.

The algorithm selects multiple expression combinations, just like using multiple sentences to describe the same thing, to help explain and detect breast cancer.

“We use multiple models that have all proven high accuracy in predicting breast cancer,” Molinari explained. “We don’t want to rely too strictly on one model since it could yield contradictory results as occurred in past studies.”

The statistical models could help find common indicators that can precisely predict breast cancer.

He is collaborating with researchers at the University of Lausanne and the University of Geneva. With data already collected and analyzed, Molinari is seeking funding to further develop this algorithm, confirm the results and apply it to different fields of research.