Abhishek Biswas is not one to shy away from a mystery. "I like detective stories," he says, "where, if you read carefully or logically, you can figure it out." A Research Software Engineer at Princeton University, Biswas is especially fond of Sherlock Holmes, the London detective who famously let no detail slip his notice. Perhaps that explains Biswas's attraction to metagenomics, which attempts to collect all of the genetic information from every microbe in an environment and then reconstruct a scenario that fits the data. "You collect a sample," he says, "and it has thousands of different bacteria at different abundance levels. They collect terabytes of data, and you have to figure out what's in that mix."

As the dedicated RSE for Princeton's Department of Molecular Biology, Biswas helps scientists get the most out of their datasets. "They're trying to tell a biological story," he says, "but they need some analytics tools. I help them develop those tools." He contributed to a research paper published in Science in December 2019 that found a new way to look at the menagerie of bacteria, viruses, fungi, and other microbes that make their home on our skin and mucosal surfaces. Working at Princeton's Donia Lab and led by postdoctoral researcher Yuki Sugimoto and graduate student Francine Camacho, the team developed computer algorithms that use metagenomic sequencing data to detect enzymes produced by the various constituents of the human microbiome. Mohamed Donia, an associate professor of molecular biology at Princeton describes the collaboration as "a remarkable and cross-disciplinary partnership between software engineers and experimental biologists." "Before Abhishek’s work, this task was simply unattainable," he says.  "But now, thanks to his remarkable skills in software engineering, we have an efficient algorithm to achieve this goal."  "They produce so much data, you can keep computing and get amazing analytics out of it," says Biswas, who developed analytic software that helped the researchers discover a class of clinically relevant molecules. "It's a huge big-data problem." Like Holmes, Biswas doesn't stick to one type of problem for too long. "I get to change between projects like imaging genomics, protein structures, viral biology," he says. "We generally try to change every nine months."

During one study at the Devenport Lab that examined how hair follicles grow, he saw that, while a microscope generates a stack of 10 images, the researchers were selecting only one frame out of each stack to analyze.  Biswas, not one to overlook evidence, developed a tool that let the scientists integrate the slides into a much fuller picture. "We went from 2D to reconstructing the shape of the cell in three dimensions as it changes as the hair follicle grows," he says. "They were able to extract a whole new biological story out of that." "We expected skin cells to be organized like bricks, neatly packed in cuboidal arrays," says Devenport. Instead, she says, Biswas's tool revealed that the cells adopt an energy-efficient shape called a "scutoid" –  think of a hexagonal prism with one of its sides sliced off at an angle – that allows cells to twist around one another. "We are now using his tool to define the 3D basis of tissue folding."

Princeton is one of a handful of universities in the United States with a centralized RSE group, a team of academic software specialists who develop tools for research projects in other departments. The concept, which owes much of its origins to a conference in the United Kingdom in 2012, aims at professionalizing research software development, a job that all too often has relied heavily on transient programmers writing homebrew code. "Every branch of science needs complicated analytics tools right now," says Biswas. "If these tools are going to be maintainable and we're not going to have graduate students reinventing the wheel, then RSE programs need to be part of the framework of any university." Having a dedicated RSE also offers the benefits of expertise without creating the pitfalls of over-specialization. "If one lab did a particular sequencing," Biswas says. "I can share the best practices to start with." As for Biswas, he's happy to have found a professional home where success isn't always measured by publication metrics. "We look at projects where my involvement would be the most useful,"  says Biswas. And even though he loves detective fiction, Biswas doesn't mind leaving the writing to others. "It is generally not the part I enjoy," he says. "The part I enjoy is developing the code."