Dr. Arash Jamshidi is an expert in computer science and biotechnology. He has researched at the University of California, Berkeley, and Simon Fraser University. As of the last decade, he has been working at Illumina and GRAIL, two companies dedicated to medical biotechnology. I was allowed to interview him about his industry and how he got to where he is today!
Could you tell me a little about how computer science and data science are applied to biotech?
Biology was based on very limited datasets until the last two decades. It was challenging and time-consuming to conduct experiments that generated precious data.
Much of this changed due to new technologies that came on board, maybe around the 1990s to the 2000s, with next-generation sequencing and many other approaches. We could look at the genome and many different aspects of biology at a much higher resolution and throughput.
These techniques started generating massive data sets to look at biology more quantitatively. The only way to do that with the scale of the datasets is through using computer science, deep analysis, and machine learning. This trajectory will continue to be there and expand in the future.
Have people been using AI to look through the data?
A lot of the initial approaches were to develop the tools and to be able to analyze the data because the size is just huge.
So if you, for example, looked at DNA sequencing, you'd start to look at the whole genome. Even analyzing and deciding on all the reads you're getting from a sequencer about where they align in the genome requires algorithmic developments. But more recently, machine learning has become much more prominent in that now you are starting to ask very different types of questions about the data.
The core of machine learning is about asking a question. The data and modeling of the data cannot give me the answer. I don't have a specific formula for answering that question. If I'm trying to, for example, categorize images of dogs and cats, I'm trying to train an algorithm solely based on the training of the data.
There are many parallels in biology that are now becoming prominent applications. For example, cancer detection is something I've worked on very closely. It's almost impossible to create one specific formula to go though the all the vast data sets generated from many different individuals' genomes. But, if you expose the algorithms to the data, they can find patterns.
Do you think using AI for genome datasets will be the future of this field?
In some aspects, there will always be a lot of dependence on creativity, a deep understanding of biology, and biological models that data can prove or disprove those hypotheses.
That will continue to be there, and there's no replacement for that creativity. But, there are many questions that you wouldn't be able to ask or conduct experiments for to determine that hypothesis unless you have access to the data and analysis. From that perspective, AI will surely be a big part of the future.
Could it ever completely replace humans in this field? For example, could AI automate finding patterns in genome data?
Today that's very far from reality. One of the unique things about applying AI in the context of biotech is it's very much a multidisciplinary exercise.
You can't have computer scientists and machine learning experts figure everything out independently. It requires a team of clinicians, biologists, and others who understand the science, working with experts at the data to do practical experiments because the number of variables in biology is just so large. You can't test everything, so designing good experiments to ask the right questions is key.
Do you mostly work in teams, considering all the aspects of your field?
Yeah, absolutely. Teamwork is part of the pleasure of being in biotech from a computational side. You get exposure to many different fields and colleagues from other areas, enriching your knowledge. So, absolutely. It's big on teamwork.
What part of your work do you focus on? For example, do you do the coding, or do you guide the coders?
I try doing as much hands-on and coding as possible in my roles. More recently, I've managed large teams of data analysts, data scientists, and machine learning engineers. A lot of my time has gone into guidance and helping design experiments, but not doing the day-to-day coding. I'm trying to gain some of that back because there's a lot of pleasure in doing your own hands-on analysis.
Could you share a general overview of your professional journey?
I am originally from Iran. I went to university there and immigrated with my family to Canada, where I finished my studies in electrical engineering, specifically in signal processing.
I then moved from Canada to the U.S. to UC Berkeley to complete my Ph.D. in electrical engineering and computer science. I was working on many different technologies related to biotech. Some of that work was later licensed by Berkeley Lights, which conducts high throughput analysis of single cells.
But, I wasn't directly part of that company. After my Ph.D. work, I became interested in the genomic space, so I moved to the company Illumina, the leader in sequencing. I was part of the research group there for about five years, working on things including high single cell analysis and extraction sequencing.
Near the end, we started working on applications in noninvasive screening and diagnosis, particularly for several types of cancer. That became part of the initial group, and the work spun out in the form of a company called GRAIL. I was part of the founding team there.
And GRAIL was focused on early cancer detection using fragments of blood DNA in a very non-invasive way. I was part of GRAIL for about six years near the end of last year. We made fantastic progress during that time. The company grew from about 40 people to almost seven or eight hundred.
We launched a product called Galleri, which is for multi-screening of many different cancer types. And the company was acquired by Illumina back about a year ago.
You mentioned that you did some work in research. Was that different from working in the industry? Which do you prefer?
The exciting thing about biotech is that the lines between research and industry are constantly blurring.
You can only do some of these extensive experiments in a company focused and with the right amount of resources. Even though I transitioned from academia to industry, I felt like I was still doing a lot of R&D and research. This is probably not the case for other fields that are more mature in terms of the types of technologies that they're working on. But, I've always found them to be very close.
If you forced me to pick one, I would say industry is more interesting, but they're very close.
Biotech is in a unique place. I don't think it often happens where the latest discoveries are not too far from the latest applications. So, for example, if you think about particle physics or theoretical physics, there are so many fundamental discoveries. However, the applications could be decades away from ever becoming a reality.
Whereas in biotech, something like sequencing wasn't even available until two decades ago, but it has already gone into so many applications, even though it's a very new technology.
How has your field changed from when you started in the industry?
It's been maturing in several ways. Maybe a decade or so ago, there was a lot more focus on developing the tools to get them to a level that can do quantitative biology at a large scale and be able to analyze the data. The community has done such an excellent job of advancing these tools that now we can take them to applications. Now, applying these tools in the context of cancer biology, neurodegenerative diseases, or pregnancy has created some excellent new applications.
Are there any downsides to your job?
Like everything else, there's always a scientific risk. Biology is unique in that you can ask fascinating questions, and there are many open problems, but it's not always clear whether we know enough to address them.
For example, one of the essential things, especially if you're more on the R&D side, is the ability to handle uncertainty because we are asking many questions in a new way. Some of these may not pan out: the biology may not be there, or the experiment may fail.
It's different from, for example, building a piece of software because it's about engineering and making it. The endpoint is clear. But in R&D, there's more uncertainty about the results. The other thing that's a little difficult about biotech is that it's time-consuming and resource-intensive to do experiments and large-scale trials.
Do you have any advice for students who want to study computer science or biotech?
I highly encourage it, and I think it's a very rewarding field. It's very fortunate to do something that you like, in terms of science or discovery, but also have the ability to put that to good use and solve significant problems in the world.
Biotech offers intellectual curiosity and is at the cutting edge of science and scientific discovery. Not everything is theoretical, and most of it is developing applications that will impact people's lives. It's a privilege to be in a position like that.
Do you have any tips for job interviews?
It's always helpful for job interviews to be as knowledgeable as possible about the company you are speaking to and what the role entails. It's good to prepare for the material they may be exploring, but don't feel shy about asking questions in the interviews to make sure it's a good fit for you.
The most precious resource anyone has is their time and their experiences. Joining a company or selecting initial positions are a part of that. I also want to ensure I'm not suggesting being too selective.
As long as you're directionally in the right area, you'll find your way through these experiences that present themselves.
In Short:
Working at the intersection of biotechnology and computer science means a lot of teamwork on the job. You'll be collaborating with specialists from all sorts of fields.
You can still research if you're working for a company! Using computer science in biotech is a novel field, and companies need researchers to discover new ways to use biotech and new applications.
Since there is so much data to be found in the human genome, AI is needed to sift through and find patterns useful for researchers to detect diseases and cancer.
Comments