Research Opportunities: Prospective students are welcome to contact me via email.

Health Outcomes and Biomarkers

Duchenne and Becker muscular dystrophies are X-linked, genetic disorders caused due to defects in the DMD gene, leading to abnormal dystrophin (missing in Duchenne muscular dystrophy [DMD] or partially functional in Becker muscular dystrophy [BMD]). Among other things, our research aims to understand

- Can we explain (and predict) why DMD persons can have widely different disease trajectories (some DMD persons can be in wheelchairs by age 12, others not until late teens or 20s) even though they share the same primary biochemical defect.

- Why do some DMD persons benefit a lot from chronic steroid treatment while others do not (as much)?

- Why do some DMD persons have adverse effects (weight gain, bone fragility, Cushingoid syndrome, mood and behaviour disturbances) quite quickly to chronic steroid use while others do not?

- Can we better understand the incredible variability in BMD persons disease trajectories? Can we model and predict?

- How do clinical outcomes data, biomarkers, and patient reported outcomes complement each other? How can we run better clinical trials in DMD and BMD to have a higher probability of success?

- How can we use biomarker data (e.g., if some biomarkers are correlated) to better understand pathophysiology?

We use clinical outcome (from clinical trials and natural history studies) and biomarkers data, both molecular (proteins, microRNAs, mRNAs, genetic modifiers) and digital (accelerometer-based) for understanding these rare diseases better.

Cluster and classification analysis (Statistical Learning)

We develop new mixture models (and software) for cluster and classification analysis. Among other things, these can help with

- Understanding which data observations are more similar to each other, and which are not. Data points similar to each other are grouped in a cluster together. For example, if these are patients being grouped together based on gene expression levels, this informs us about similarities in a cluster and differences between clusters. This has implications for disease subclass identification, prognosis categorization, treatment response, etc.

- Grouping data without any prior knowledge of which data points belong together OR when there is some prior knowledge on grouping of a subset of data points.

Phylogenetic Comparative Methods

We develop new phylogenetic models that allow us to test evolutionary hypotheses given a phylogenetic tree. Among other things, these have involved

- Developing phylogenetic models and software packages for modeling evolutionary rates while accounting for missing data (due to degradation of extant DNA or a mass pseudogenization event)