Under a full sturgeon moon, Code Ocean recently gathered together at Catalyst Restaurant near the MIT Campus in Cambridge, MA with a group of some of the best minds in computational biology to discuss an issue of great importance to all of us: How to manage science teams at scale. It was a beautiful evening, and though Catalyst never disappoints when it comes to food and drink, the discussions took their rightful place as the real stars of the event.
After cocktails, a panel of executive speakers shared their personal experiences in scaling computational research teams. The consensus was that, at this point in time, computational research has become too complex for one individual to manage. It requires a robust team of people who interact multiple times daily.
Some industry experts at the event pointed out that the new generation of scientists is full of hybrid creatures who possess a combination of deep scientific understanding, statistical knowledge, and coding skills, capable of both asking difficult questions and building robust hypotheses through data analysis. This triggered an interesting debate between two approaches to team growth – the first focused on hiring hybrid scientists who understand both biology and coding and the second approach on scaling pods of coders and non-coders in tandem.
Regardless of the approach to hiring, the fact is that building a robust team is no longer enough, in and of itself. Even if you manage to build a team of specialists with the kind of hybrid talent you are looking for, leaving each team to its own devices to manage practices will only lead to chaos. In a field as complex as computational biology, the path to success requires taking a cross-disciplinary approach, as well as observing best practices in administration, management, and collaboration. In order for everything to work smoothly, leadership practices need to be supported with an advanced cloud-based technology framework that enables seamless, continual sharing of data and code.
An up-front focus on shareability requires a short-term investment but provides long-term gains in the form of enhanced consistency and quality, team efficiency, and accelerated pace of results. As explained by Yair Benita of AION Labs, implementing a cloud-based platform such as Code Ocean’s inherently creates standards in data management, coding practices, and documentation that enable efficient data sharing and collaboration within the organization. Code Ocean’s platform allows teams to develop practices that empower team members to learn from one another, reuse code efficiently, and work together to solve the biological problem at hand, rather than spending time trying to get things to run.
Benita offers a list of questions team leaders should ask themselves when setting up their research teams:
- Team management: How do you ensure analysis standards across your team? Even from the most common tasks like calling DNA mutations, or comparing gene expression of two groups or even simple tasks such as computing a drug response curve. Is there a team standard for the group that is assigned by your top expert? Is the core code maintained and improved over time? Can you track who did what and in which analysis with which data and parameters?
- Data management: When you have a big team, you may have large volumes of data and results to track. Is everyone using the same packages, and seeing the most recent version of the data? How can you capture, sort, track, and use it in a way that is shareable?
- Collaboration: Can team members learn easily from one another? Can people easily find analyses others have done? Can code and analysis be reused? Can results be shared with non-coders? Can simple analysis be deployed throughout the organization as applications without significant engineering resources?
Giving some thought to these questions in advance will save time and ensure better-quality results in the long run. By observing leadership best practices and supporting them with advanced technology, you will enable your people to do their best work.