First person – Huy Tran Open Access

Huy Tran

Describe your scientific journey and your current research focus

I received my formal training in signal processing and biophysics at Tampere University of Technology (Finland), where I pursued my PhD on the dynamics of gene expression in bacteria. During my first postdoctoral position at Institut Curie and École Normale Supérieure (Paris, France), I extended this research to early embryonic development in fruit flies. In 2022, I returned to Finland for a second postdoctoral stage at Tampere University, focusing on ionic signaling in epithelial tissues. My research has always been at the very interface between biology and physics, where I have assumed each of all dry-lab works, from creating imaging analysis toolboxes to statistical analysis and mathematical modelling. This allows me to be closely involved in the design of quantitative experiments. In 2024, I was awarded a Research Fellowship by the Research Council of Finland and launched my own research group. We are currently investigating how ionic signaling interacts with gene expression regulation to predict the long-term behavior of tissues across various biological contexts.

Who or what inspired you to become a scientist?

Probably not a single person or phenomenon in particular. I began to see a clear future in science during my first postdoc in Paris, France. I was very much immersed in the vibrant research community, with eminent biophysicists, biologists, evolutionary geneticists, all reachable within a couple of city blocks of the 4th arrondissement. I am convinced that, despite the inherent heterogeneities, stochasticity and even erraticism, living systems are tailored over billions of years of evolution to ensure the organisms’ survival. Therefore, they must operate either in equilibrium with their environments or at the physical limit. This idea continues to fascinate me and drives my pursuit to better understand the principles behind life's complexity.

How would you explain the main finding of your paper?

In this work, we develop the software toolbox EyeHex (Eye+Hexagon) to segment ommatidia (i.e. facets) from fruit flies’ 2D images. It features two integrated modules: the first utilizes machine learning to preprocess different types of images for easier detection of the eye and ommatidia areas. The second module leverages the hexagonal organization of the compound eye to locate individual ommatidia. EyeHex offers a cost-effective, rapid, and flexible pipeline for extracting detailed statistical data on Drosophila compound eye variation, making it a valuable resource for high-throughput studies.

What are the potential implications of this finding for your field of research?

Variation in Drosophila compound eye is a broadly used model in evolutionary developmental biology. However, the required manual segmentation of ommatidia in compound eyes is extremely time consuming and labour intensive, given the high number of ommatidia, often ranging from hundreds to over ten thousand per eye. This challenge has relegated studies of phenomena like fluctuating asymmetry in Drosophila to simpler traits (e.g. wing morphology and mechanosensory hair patterns). EyeHex bypasses this bottleneck by enabling accurate segmentation and extraction of eye morphology statistics across imaging platforms, including accessible brightfield macroscopy. This will facilitate larger-scale studies to detect compound eye size and morphological variations, typically required to address fundamental evo-devo questions.

Which part of this research project was the most rewarding?

It is the evolution of our analysis strategy during the course of this project. Originally, we thought Artificial Intelligence would make the task low-hanging fruits, since we human identify ommatidia in the insect compound eyes so effortlessly. It turned out that, unless we had thousands of training eye images (which no one would want to segment manually), the classification of ommatidia regions from machine learning models is rigged with noise. Through brainstorming, we proposed to map ommatidia one-by-one to an elastic hexagonal grid. This both overcomes a noise problem and preserves the apparent hexagonal organization of ommatidia in the compound eyes.

What do you enjoy most about being an early-career researcher?

Pursuing science was worry-free! This is of course furthest from the truth as we progress further in academia.

I appreciate the time when we were allowed to be naïve, be curious and see possibilities in everything. We saw friends and companions in fellow scientists, exploring questions and hypotheses from serious to inane. Moments like those are now rarer, as we become more independent and have more responsibilities, but enough to remind us of such privilege we have been enjoying.

What piece of advice would you give to the next generation of researchers?

One of the most important skills for a researcher is posing relevant questions during interactions either with fellow scientists, with your students, or with your grandma. It is indeed a combination of skills, including information recapitulation, identification of knowledge gap, and communication. Learning to pose questions remains crucial in setting us apart, given how much society has been increasingly delegating this skill to ChatGPT and the like.

What's next for you?

Survive academia! I recently started a fledging group on quantitative cellular communication, with my own funding and students in Tampere University. The whole learning curve as a PI is quite daunting. It is exciting to see that my students can now carry out experiments themselves and that we are not on each other's feet all the time.