We develop computational methods that integrate genomics, molecular evolution, and machine learning to uncover novel mechanisms by which infectious agents interact with host processes to drive disease.
We are located at The Wistar Institute.
Active research programs
Tissue-Resident Microbes and Their Role in Disease Modulation
Our lab develops machine learning and sequence analysis frameworks to identify and characterize microbes that influence human disease, with a focus on cancer and autoimmune disorders. We are building models to detect microbial reads, designing computational methods to quantify microbial expression in human tissues, and conducting integrative analyses to link tissue-resident microbiota to patient outcomes.
Quantification of Gut Microbial Genes and Development of Fecal Biomarkers
The gut microbiome holds great promise for advancing cancer clinical decision-making through non-invasive fecal biomarkers and fecal microbiota transplantation. Our lab develops computational methods to improve the quantification of microbial genes and proteins in fecal samples. These methods enable more reliable and reproducible biomarkers for predicting clinical outcomes and provide deeper mechanistic insights into the immunogenic functions of the microbiome.
Detection of Pathogenic or Immunogenic Viral Sequences
The human virome is highly heterogenous with numerous uncharacterized members. Our lab develops integrative machine learning and sequences analysis methods to identify new viruses, viral genes, or genomic regions that are associated with disease risk, severity, and immune responses.
Biologically Informed Classifiers of Cancer Treatment Responses
A major barrier to the clinical translation of machine learning in oncology is the limited biological interpretability of many models. Our lab addresses this challenge by developing biologically informed classifiers of cancer outcomes, integrating prior knowledge from biological networks and databases with cancer omics data to improve both predictive power and interpretability.