Return to site

A Cancer Gene Learning Machine

The Horizon High School Research Projects

The smallest variation in a person’s genes, a single nucleotide polymorphism (SNP), is just a switch of a nitrogenous base — perhaps an adenine goes where a cytosine normally sits. Most SNPs, or “snips” as they are colloquially known, are harmless and reflect normal human genetic diversity, but some can cause cancer, cystic fibrosis, and a host of other diseases. These microscopic differences are what Karthik Mittal set out to study with the Horizon Labs Research Program, using the extraordinary power of machine learning to review genetic data.

A headshot of Karthik Mittal in a blue oxford and blazer. Karthik recently completed a research project with the Horizon Labs research program, an extracurricular academic and mentorship opportunity that aids students in conducting their own investigations into different aspects of how our world works. Photo courtesy of Karthik Mittal.

Karthik Mittal

Karthik, a rising high school junior from California, is a young man of many talents; in addition to taking four AP science and humanities classes next year, and when he’s not devoting his free time to academic enrichment research projects with Horizon, he loves ceramics, the guitar, and battling his brother Aditya in tennis. But his greatest passion, stoked from an early age, is in data science. As Karthik recalled, “I took a class when I was seven years old and fell in love… It was basic Python, but it intrigued me so much, solving problems with a computer.” With Horizon Labs’ high school research program, Karthik had the chance to funnel that enthusiasm into a research project with a potentially big impact.

Karthik researched SNPs associated with breast cancer — the free MRBase library provided a slew of genome wide association studies to analyze. Without any funding, Karthik relied upon the MRBase library, to which he was introduced by his Horizon Labs mentor Dr. Parsa Akbari, a researcher at the Department of Statistical Genetics at the University of Cambridge, to provide the raw data for his research. He used two unsupervised machine learning techniques, K-Means and Hierarchical Clustering, to find patterns in the SNPs, ultimately demonstrating that the MAPK signaling pathway is key in breast cancer development and that the genes ESR1 (Estrogen Receptor 1) and BRCA1 (Breast Cancer Gene 1) are linked.

Karthik puts together his research paper and runs data analysis at his workspace. Karthik used the MRBase to obtain genetic information, a source he was turned on to by his Horizon Labs academic research program mentor for his high school research project. Photo Courtesy of Karthik Mittal.

Karthik at Work

He discussed his process in an interview testimonial with Horizon research associate Zak Rosen, transcript below.


Zak Rosen: Hello Karthik! What led you to get involved with Horizon?

KM: I wanted a capstone project and started looking around. Horizon really suited my needs. It seemed like a great opportunity to… expand my horizons!

ZR: Good one. I have to ask… What’s more fun, K-Means or Hierarchical Clustering?

One of Karthik's Hierarchical Clustering. This diagram includes the SNPs on the x-axis and the associated distance between them on the y-axis. Photo Courtesy of Karthik Mittal and the Horizon Labs Research Program,  which organized Karthik's high school research project.

Karthik's Hierarchical Clustering of Carcinogenic SNPs

KM: K-Means. The dendrograms produced by Hierarchical Clustering are boring, K-Means has greater visual appeal. That’s just my preference though, they’re both fun to use.

ZR: What’s the next step in your research?

KM: I want to explore other unsupervised machine learning algorithms. I’d like to expand my research into other cancers. And I’d love to get some funding to do test trials, see if my findings produce viable results. With the MAPK signaling pathway, I’m basing my findings on other people’s research… with funding I’d be able to do my own.

ZR: I know one of the coordinators for the program is helping you publish. How does that feel?

KM: I’m really excited to get into a scientific journal. It’s an opportunity to showcase my work to the entire world. Horizon has invested a lot into helping me through the process.

The heading of Karthik's research paper, a sneak people of what people will be seeing when he gets published. Karthik undertook the project with the help of Horizon Labs, an academic program for gifted high school students that lets them undertake independent and guided research.

ZR: Can you talk more about that?

KM: My mentor, Dr. Akbari, was really helpful. For first ten weeks I learned how machine learning ties into genetics. Then for three weeks I had my project. Whenever I had questions I would email Dr. Akbari. His vast array of knowledge was incredibly helpful. I learned more about machine learning, I learned how to write a research paper, I learned about biology more broadly…

ZR: Your brother Aditya also worked with Horizon Labs. What was it like working side by side?

KM: Competitive. We always fought with each other to see who could get more pages, whose our mentor would like more - this pushed me to produce higher quality work.

ZR: Has this helped you think about what you want from your future?

KM: KM: Definitely. There are so many applications that suit my joint interest in machine learning and biotech - Horizon has taught me how fun it is to pursue research. This will be really helpful with college and my PhD.

ZR: You want to get a PhD?

KM: That’s the plan.

ZR: How do you feel about everything you’ve accomplished?

KM: Writing a research paper in four weeks is extremely challenging, but I’m really glad I chose Horizon, they really helped me through it. I want to pursue more research in this field.

ZR: That’s wonderful to hear. Thank you so much for this interview Karthik, it was great meeting you.

KM: You’re welcome! Thank you for the opportunity, it was really fun talking to you!

All Posts

Almost done…

We just sent you an email. Please click the link in the email to confirm your subscription!