Sparse neural network for modality selection of multi-omic data
In this master's thesis, we will develop and utilize a novel sparse deep learning framework designed to perform simultaneous prediction and modality selection applied to microbiome and metabolite datasets. You can read the full proposal here
.

Predicting the effect of drug perturbations on the human gut microbiome
This project aims to develop predictive models that can assess a compound's impact on the human gut microbiome. You can read the full proposal here
.

Proteome based bacterial representations
By reducing embedding redundancy and incorporating supervised fine-tuning, we aim to produce efficient protein representations for a better microbial representation to enhance downstream analyses. You can read the full proposal here
.
“Biomedical Statistics and Data Science”