#WhyIScience Q&A: A computational biologist uses physics to find hidden patterns in cells

Niranj Chandrasekaran talks about his work developing a large image-based cell profiling dataset and his long-standing fascination with physics and astronomy.

Portrait of Niranj Chandrasekaran
Credit: Allison Colorado, Broad Communications
Niranj Chandrasekaran

Niranj Chandrasekaran has always loved the night sky. Growing up, he spent nights on the terrace outside his family’s house in Chennai, India with his dad, a software engineer who’d studied physics. Together, they searched for meteorites and stared up at constellations, looking for patterns among the stars.

Today, Chandrasekaran is still looking for patterns, this time in biological imaging data. After receiving his PhD in biophysics from the University of Chapel Hill at North Carolina, he joined the Broad Institute of MIT and Harvard as a postdoctoral associate in the Imaging Platform. There, he led the Joint Undertaking in Morphological Profiling with Cell Painting (JUMP-CP) consortium in creating the JUMP-CP database — a public reference collection of cell images generated with the Cell Painting microscopy assay to accelerate drug discovery. Developed by Anne Carpenter, senior director of the Imaging Platform at the Broad and others, Cell Painting reveals subtle features in cell images, turning biological traits into quantifiable variables that reveal the effect of turning genes on or off, or treating cells with a drug.

The dataset, which was released in November 2022, contains information from more than 2 billion cells and 140,000 samples. Chandrasekaran says it was the first publicly available dataset of its kind. He’s been involved in every stage of the process, including managing the 10 pharmaceutical companies and two nonprofit organizations contributing to the database, designing experiments, and creating data analysis pipelines.

We spoke with Chandrasekaran about his early interest in the physical sciences and his favorite parts of the JUMP project in this #WhyIScience Q&A.


When did you first become interested in research?

I've always been interested in basic science and asking questions about how things work. Because of my interest in astronomy, I was initially drawn to the physical sciences. But growing up, science was something that you studied at school. I never thought I’d be able to do research on my own. Then I read a book about Albert Einstein and the theory of relativity. The idea of finding something new, no matter how small, got me interested in research. 

During undergrad, I worked in industrial biotechnology, which is a combination of biology and physical sciences, so I was able to focus on doing biology research and scratch my itch for the physical sciences. When I came to the US to do my PhD, I shifted into molecular biophysics and studied how proteins wiggle inside the cell. Studying protein dynamics, you use the same kinds of equations and physics that you use to study planetary motion. I felt like a physicist.

Afterwards, I had a brief foray into the world of genomics, and then I found image-based profiling at the Broad. I saw that all the coding and data analysis skills that I gained during my PhD translated into this space. I've shifted between a lot of fields within biology, which can be challenging, but I've always taken the experience I've gained from one of those fields and applied them to another. I think that’s given me a unique perspective. 

What is image-based profiling and what kind of impact do you hope it will have?

Image-based profiling involves converting biological images into numbers and finding patterns in those numbers. We perturb cells — add chemical compounds to them, or knock out or over-express genes — and then study the impact of these changes on the cell by taking images of them.

This methodology is gaining popularity, especially in pharmaceutical companies and biotechs who are interested in using it to figure out how chemical components work in a high throughput manner. It can also be used for screening compound libraries to identify potential drug candidates and assessing toxicity of drug candidates. Nowadays, drug discovery ends up being an extremely costly process because the research required at the initial discovery phase can be very time-consuming. Any way we can accelerate that process will help bring down the cost.

What does an average day look like for you?

My work has changed quite a lot over the last three years because in the beginning phase of the project, we were planning experiments. Now that the dataset has been generated, we’re focusing completely on data analysis. This is my bread and butter. I’d been waiting almost three years to get the data and immerse myself in it, trying to find interesting patterns. Data analysis can be frustrating when the numbers don’t make sense, but when I’m reminded of the translational impact my work could have and how it could help someone who is going through something difficult, I get motivated.

What advice do you have for young scientists interested in biology?

Don’t get bogged down by the pressure to decide which direction your career should go when you're a teenager. It's okay to take your time. I've been able to shift through so many fields. As long as you’re in an environment that encourages active learning, you'll be able to use the skills that you've gained and get up to speed. It's totally fine to explore, and you never know what fields are right around the corner. Image-based profiling didn't exist when I was an undergrad. I never could have predicted this when I was younger.

How do you hope this field will develop in the next 10 years? 

The progress we've had in image-based profiling is in part due to advancements that have been happening in computer vision, where people have been coming up with new methods to extract information from images. In the past, when people invented new things, it might take a few years before it was used in our field. But now, whenever something new comes in computer vision, people adapt it immediately. I’m looking forward to the next decade and the advancements in computer vision and deep learning and how they’ll impact biology.

I think this is the right time to be in this field. We’ve started generating large data sets. And people are going to start using these data sets and taking methods from other fields and making discoveries. In the past 10 years, we’ve overcome a lot of the problems any nascent field would have. Now, we can reap the rewards and make impactful discoveries.