Anatomy of Canada’s Largest Bioinformatics Hackathon
On the third weekend of September, the Donnelly Centre hosted the second-ever Toronto Bioinformatics Hackathon (TBH): a two-day marathon of learning as programming enthusiasts team up to answer novel research questions and develop functional software projects. This hackathon is centred on computational biology, the interdisciplinary application of methods from computer science, statistics, and mathematics, to address biological questions.
Hackathons challenge groups to balance time management, skill, teamwork, and sleep. This is what that balancing act looks like at all levels, from the winning projects to the private judging deliberations.
THE COUNTDOWN BEGINS
As the clock hits 9:00 a.m., the timer officially starts. Participants, informally called “hackers” mark the start of their project with their first GitHub commit. The teams form in circles to begin the first of many hours of work, with the more seasoned groups gathering around tables piled with snacks.
“This is my first time leading a hackathon team,” says Dennis Zhu, University of Toronto undergraduate and lead of a project using machine learning to predict enzyme activity. “Organizing the project is a big challenge because of how complex both the biological and computational sides are.”
The large meeting table in Donnelly's White Room offers cans of soda, water, and energy drinks—the latter of which already has the biggest dent in the supply. Everything in this hackathon, from entry, to meals, to lodging, is provided for free.
“Is this okay or do you want to work somewhere else?”
“It’s okay.”
“Pull up a chair. We’re having issues with uploading the files...”
There is a point when the planning and discussions dissolve into silence. In the hour leading up to lunch, the most prevalent sound is an incessant clicking that can be heard throughout the second and sixth floor. It’s the unending typing sounds of over a hundred hackers creating, then slowly unravelling, a Gordian knot of spaghetti code.
The timer has started. By the time the organizers call for a lunch break, four hours have already elapsed.
“I think preparation is key,” says Pamela “Pam” Alamilla, the team lead of COMPILE, a knowledge graphing project working to visually represent protein interactions. “We're extracting information from literature, and that’s a lot harder than we anticipated. It took us a little while to get the data in order.”
Following the success of the Human Genome Project in 2003, current applications of computation biology often involve extracting insights and information from publicly available data sets. It’s a collectivist approach to knowledge, according to Jacob Fine, founder and operations director of the Toronto Bioinformatics Hackathon.
“Petabytes of biological data are freely accessible,” says Fine, a computational biology PhD candidate. “The real ingenuity is figuring out how to integrate the data, model it, and extract meaningful biological insights. That’s where our participants’ talent comes in.”
The word hackathon, deriving from “hacking marathon,” highlights the endurance needed for teams to cross the finish line. The experience itself is unlike anything else, says judge and TBH co-founder Elia Afanasiev, a computational research technician at the Goeva Laboratory at Donnelly.
“It’s fun to have a dedicated time to just do something, anything—talk to people, pick up a new skill,” says Afanasiev. “It’s project driven, which for programming—at least in my experience—is the best way to learn.”
While the hackathon took place over the weekend, two months prior to the date, the organizers opened submissions for project proposals to ensure the projects were scientifically rigorous. This is a contrast to regular hackathons, which usually give their project themes day-of.
Afanasiev says that defining the project ahead of time is for the participants’ benefit, “while it’s quite common for teams to pivot entirely, they tend to stitch something together.”
Teams this year are developing a wide variety of projects: everything from a distilled version of Google’s AlphaFold model, to a tool for detecting Alzheimer’s disease from speech recordings. Because one weekend isn’t enough time to develop a polished product, the teams will be judged on proof-of-concepts, incomplete models, or models that do not account for edge-cases.
The limited amount of time, however, is a feature and not a bug for Dr. Jeff Wintersinger, event judge and Principal Scientist at the biopharmaceutical company Deep Genomics.
“Most things that I’ve learned that have benefitted me as a scientist have been entirely motivated by ‘I have this research idea, I need to pursue it, and I’ll learn whatever I need in service of that goal,’” Dr. Wintersinger says. “Compressing that timescale, so you’re collaborating on a short time frame, helps people understand what’s important.”
Teams continued that collaboration late into the evening, during the optional overnight lodgings at the Bahen Centre for Information Technology. Fuelled by catered Italian food and a midnight snack run, participants hit a second wind around one in the morning.
It wasn’t all hacking though; there were games, conversations, and intermittent bouts of much needed sleep.
“At 3:00am, a bunch of us got together and started to talk about science. Everyone was so tired, so everything was so natural,” says TBH co-founder and programs director Purav Gupta, a bioinformatics undergraduate student at the University of Toronto. “These are the small things I enjoy, because you’d never do in your real life, but in a Hackathon it’s so normalized.”
Some teams worked through the night, and some recharged through sleep. Come morning, however, all were racing against the same 1:00 p.m. submission deadline.
FIVE TABLES: THE JUDGING PROCESS
Following the last moments of submission, teams clear out onto the second floor of the Donnelly Centre for lunch, inhaling the provided submarine sandwiches with efficiency. Everyone dons a bright blue Toronto Bioinformatics Hackathon t-shirt, heading to the Donnelly atrium for this year’s group picture.
For a variety of reasons, not every team submits their project: from losing work that wasn’t backed up, to technical difficulties, to not having enough time to adequately address bugs. All teams that were informally interviewed, wishing to remain anonymous, conclude that the experience itself was nonetheless fantastic for learning and motivation.
For the teams that did submit, judging occurs in two parts: the public presentations and the private deliberations.
In Donnelly’s Red Room, five tables are labelled A to E and set up in a loose circle, surrounded by canary-yellow lecture chairs. The teams present twice, to two different groups of judges. Each presentation rotation flashes by in eight minutes, with five minutes of questions and clarifications. Afanasiev calls out timing warnings for the teams.
“Thirty seconds until the next rotation. Judges get ready.”
One of these judges is Dr. Aleksandrina Goeva, an Assistant Professor in Computational Biology based in the Donnelly Centre. While she looks at the product and presentation, she’s also paying close attention to the experience of the teams.
“Did it seem like the group felt everybody was integrated—that everyone’s contribution was meaningful, versus an imbalance?” Dr. Goeva asks. “I think they should be motivated by the experience and the knowledge they might get. They are going to spend two days here, so they should get something out of it, whether they win or not.”
Dr. Wintersinger echoes the sentiment, saying he’s interested in whether teams cohered as a collective. He adds, "I look at if they have a coherent outlook on what they can do to condense [their project] down to something they can develop in a short time frame.”
One of his favourite questions to ask participants concerns how they can build upon their hackathon work.
“If you had fifty grand, or fifty more hours, how would you further the project?”
“Um... I can give my thoughts and you guys can expand on it—My initial inclination would be to actually fine tune the models...”
BEHIND THE CURTAIN: HOW HACKATHON WINNERS ARE CHOSEN
The judging table mirrors many of the team setups this weekend: there are coffee cups, water bottles, diet sodas, and a chip bag amongst the scrawled-upon rubric sheets.
The private judging takes place afterwards in Room 612 at Donnelly. The deliberation is an extension of the democratized thought the judges are looking for in the teams: they sit around an oval table and share their thoughts without talking over each other.
“I was really surprised—there was no project that didn’t use deep learning or AI.”
“These guys didn’t.”
“Oh, really?”
“They talked about doing it eventually.”
“They didn’t succumb to the temptation?”
They decide on one winner by a unanimous response to, “Can I get a yea?”
Another winner comes down to a nostalgic heads-down-hands-up vote.
They evaluate the projects on the following areas of product design: interdisciplinary integration, novelty, usability, safety considerations, and qualitative change to the field. The final judgements don’t just cover the quality of the work, but the quality of the teamwork itself; rewarding democratized roles distribution and inter-team mentorship between senior project members and less experienced teammates.
“I don’t know if they actually produced in the end.”
“What matters more, what they produce or their perseverance and how much they learn? Is it the product we reward or the learning experience?”
“I think the product matters.”
Last year, the hackathon had twelve groups; this year it ballooned to twice the size, with 151 participants grouped into 25 teams vying for the winning position.
“How much we’ve grown shows that Toronto is a global hub of computational biology, especially the Donnelly Centre,” says Fine. “Part of our mission in founding this hackathon is realizing all this intellectual capital that exists in this area."
As they deliberate, Afanasiev stands near a whiteboard, writing the shortlist in blue marker. Eventually, he circles the winners.
“It’s incredible to see what people can do over the course of a weekend,” says Dr. Wintersinger, brought onto the judging panel to weigh in as an industry expert. “Seeing teams coalesce so quickly on some amazing research ideas makes me recognize just how skilled and talented the current crop of researchers is. I’m glad this opportunity is here to build a bridge between industry and academia."
“Our code would not have been possible had we not met people from the industry” says Jessica Anirisaihan, co-lead of the runner up project. "Their advice really let the sails billow.”
Alamilla, who took home the MVP award, says, "Hackathons are amazing opportunities to test your skills and teamwork—invaluable for giving people the opportunity to try a new project."
The Donnelly Centre was beyond happy to help build Toronto’s ever-growing computational biology community by serving as the background of the event.
“Interdisciplinary research and collaboration are integral to our mandate,” says Stephane Angers, Professor and Director of the Donnelly Centre. “Hosting the Hackathon creates a space for bright minds to investigate the same types of computational biology questions our research teams explore every day here at Donnelly.”
As the teams and organization committee began packing up to go home, there were already whispers about what they would do next year.
WINNERS OF THE 2025 TORONTO BIOINFORMATICS HACKATHON
FIRST PLACE WINNER ($1,500): "Guardin: Federated Learning Starter Kit"
Team (left to right): Chong Wan, Oscar Heath (co-lead), Kai Lo (on phone screen), Alexander Leonardos (co-lead), and Sahil Basra.
RUNNER UP ($500): "Predicting Alternative Polyadenylation Site Choice from mRNA Sequences"
Team (left to right): Nicole Jiang, Akshita Sharma, Jessica Anirisaihan (co-lead), Boris Kafidov (co-lead), and Angelina Fernandes.
DONNELLY INNOVATION AND COMMERCIALIZATION AWARD ($1,500): "PETase Activity Prediction Using Knowledge Graphs and Graph Neural Networks"
Team (from left to right): Jared Crocco, Thomas Quigley, Denis Rivard, and Dennis Zhu (lead).
Most Valuable Player "MVP" Award ($200): Pamela Alamilla, team lead of "COMPILE: Context-aware Mapping of Protein Interactions from Literature Evidence"
Team: Jin Gong, Anushka Patel, Abhinn Kaushik, Emmy Dinh, and Syed Tabish Rahman.
