Trainee Hub

Our work is fundamentally computational, built around a thermodynamical model of gene expression similar to the Ising model from condensed matter physics. The core methodology involves fitting this model to experimental data through parameter optimization.

Specifically, we identify transcription factor binding sites and characterize their regulatory roles by maximizing the likelihood of the observed data given our model. This optimization process involves Monte Carlo simulations to efficiently explore the parameter space.

To validate our approach, we first tested the method’s accuracy on simulated data with known parameters. After confirming its reliability, we applied the methodology to experimental data from prostate cancer cells, successfully identifying functional regulatory sites within specific DNA regions.

Yes, there are several exciting applications of our work, particularly in medicine. Since diseases like cancer often result from gene misregulation caused by abnormal transcription factor activity, our methodology can offer valuable insights for therapeutic development.

By precisely identifying which transcription factors are critical within specific DNA sequences, our work can help researchers better understand the molecular mechanisms underlying genomic diseases. This improved understanding could ultimately guide the development of targeted therapies that correct transcription factor binding abnormalities or compensate for their effects on gene expression.

Our next step is to investigate how the transcription factors interact with each other. While our current work finds individual regulatory sites and their roles, we’re now developing methods to map the key gene regulatory network within specific genomic regions. This will reveal how multiple factors work together to control gene expression.

The biggest challenge we faced was a technical one related to model selection. When we fit our model, we don’t know how many transcription factor binding sites exist within a given DNA region, yet this number is a crucial parameter in our model.

To overcome this obstacle, we developed a systematic approach where we fit multiple models, each assuming different numbers of sites within a reasonable range. We then created a method to integrate information across these various models, which enabled us to determine the correct number of active sites in a specific genomic region.

Beyond the potential medical impact of our work, I found the peer review process particularly rewarding. When we initially submitted our paper, reviewers raised substantive concerns that required us to conduct additional analyses and make significant revisions.

While challenging, this feedback ultimately improved our work considerably. When we resubmitted the paper, seeing reviewers express complete satisfaction with the revised version was incredibly fulfilling. Their critical evaluation pushed us to strengthen our methodology and presentation, resulting in a much better version of the paper.

I learned that perfection can be the enemy of progress when writing scientific papers. Rather than trying to complete each section perfectly before moving on, it’s better to draft everything to a good standard first, then go back and improve all sections. It’s like sketching a drawing—you start with the basic outline and gradually add details throughout the whole canvas.

As for submission, I’ve learned to appreciate referee comments, even critical ones. Though it can be disappointing to receive negative feedback, these comments can lead to new ideas that significantly improve the work. The peer review process ultimately made our research stronger.

What kept me motivated was understanding that research naturally has ups and downs. Knowing it’s completely normal to hit slow periods, especially at the beginning of a project, helped me stay positive and persistent.