Jeff joined Fulcrum Genomics in January 2023. Jeff's career focus has been democratizing large scale data analytics. He is a member of the leadership teams for both Workflow Description Language (WDL) and the Common Workflow Language (CWL). We recently sat down to chat about work and life as a bioinformatics consultant.
What’s your area of expertise, and what excites you about your work?
Jeff: I bring a long experience working alongside the bioinformatics world as it’s evolved, literally from the microarray days to now. This has given all kinds of different experiences, from working directly with wet lab scientists in order to realize what they’re trying to do computationally, to working with computational scientists helping them along on their journey making more robust software, to developing large-scale applications for those crowds, eventually leading those efforts, and a foray out of the bioinformatics world into the Real World Data world with its heavy compliance issues and the FDA, etc.
So it’s both having seen the journey of the bioinformatics landscape and the wisdom that comes along with it - “we’ve kinda seen this before, it looks different now but…” - combined with the large variety of experiences taking the software into different places - the whole spectrum from bench scientists all the way to pure engineering teams that happen to be adjacent to science. So I fit in wherever we need the wisdom of experience in these multiple dimensions to accelerate whatever it is our clients are doing.
The snarky take on this is when you realize you’ve become your parents:
”You shouldn’t do this.”
”Shut up old man!”
”No really, you shouldn’t do this.”
”Times have changed”
But no. They actually haven’t.So it’s validating to have the ability to just not get into certain classes of problems in the first place, because you’ve done it four times before and regretted it each time!! It’s kinda nice really - “Ah ha ha! I didn’t fall for THAT again!”
What’s a common challenge in our industry that people don’t talk about enough?
The fastest way forward is knowing when to pause and build.
There can be a lack of appreciation from the scientific and medical community about what goes into the craft of good software, and how important some of the key tech details can be. It’s easy to undersell the important challenges. Another angle on this is the frequent exhortation to check in with your bioinformatician before launching the experiment to ensure statistical rigor before you spend your budget. People tend to jump in without thinking through how to set up their experiment, their computational infrastructure, their data storage regime, metadata management…
Some of this is ‘typical startup problems’ where you don’t have time to worry about this because you might not exist tomorrow. But there’s a trick to seeing the inflection points in real time and knowing when to stop and build. You need someone willing to say “Go eat your vegetables!” when it’s time. Companies can be really bad at understanding when they’ve hit one of those moments.
There’s the other side of course, to the ‘becoming your parents’ theme, because sometimes you can say “really, you don’t need this, and you will not need it for another five years. Do not worry about it now.” So again, the same problem to right-size the type and the amount of your effort.
What’s one tool, tip, or mindset shift that has made a big impact in your work?
The absolute biggest thing for me, and it helped me before Fulcrum and after joining I quickly realized it’s a huge benefit in a world like ours, was a shift from heads-down work to zooming out to see an organization at higher level scale and see the larger picture.
If you’re on team X, that processes your widgets, you get very hyper-focused on your own widget-world. You might see a problem and think that it’s the biggest problem in the company, right there in your daily tasks, and feel it must be fixed immediately. Whatever it may be. What can be hard to spot, in the grand scheme of the entire company, for example, that maybe it’s number 9,000 in the list of things that are causing problems. Maybe there are entire regions of problems more urgent to fix than yours.
Getting to have the experience of seeing larger and larger pictures, taking off the blinders for just your sub-domain, and thinking about all the challenges holistically is what will ensure that you work not on what people say they want, but what they actually need. Everything is systems-thinking, but the scope of the system can be larger.
If you could give biotech startups one piece of advice, what would it be?
The sweet spot is planning for the problems you’re actually going to have.
Before diving in think deeply about your computational data needs, data organization, metadata storage and organization - but not too deeply! Don’t become myopic around it but do try to prognosticate out a few years, and be honest about what a few years means.
”We’re going to have 10 billion vcfs!”
”No, you’re not. You don’t need to plan for that. But you may have a thousand, and what does that look like?”Think deeply enough that you can find the middle ground where you’re probably not in trouble, but you’re not over-planning either.
What’s something outside of work that inspires how you think about problem-solving?
One of the things that has always fascinated me (mostly reading up instead of hands-on because there’s not much practical utility to it) is learning about all the different programming languages in the world, particularly the small minor ones that few people ever look at. They usually have some pretty wacky ideas about how to do things. “We solve a common problem via this new paradigm” and you think “What in the world!?” And that’s why they’re not of much practical utility, but again it’s that same sort of pattern matching we’ve been talking about today.
When you’ve seen the strange corners of the academic programming world that handle a class of problem in a certain way, you can channel some of that thinking back to your work. You think of it kinda like a keynote talk at a conference. They’re not always the most useful but they do inspire a different way to think and it gets the wheels turning a little bit. So this hobby, even if not directly useful, helps to keep me thinking creatively. There was actually a conference I use to go to, I went a few times, and really liked, and one of the conference tracks was talks of exactly this nature. “This is not useful (for anyone!) but it’s kinda cool”.
So adding to the repertoire of solutions, from any domain, feeds these ideas. I’m old and I’ve seen things and the reality is that there are certain patterns that always erupt in nature and life and the more you can add to your classifier to identify patterns the better. As an example, the whole world of design patterns in software literally came out of the architecture world. There was a book in the 70s by a Christopher [Alexander] about design language for building architectural buildings, and it explored ways to talk about building patterns. Those ideas came through to software. This notion that pattern languages can be real is, itself, real. It’s all not too dissimilar to the modern AI chat bots, when you think about it. It’s decomposing pieces and then trying to spot patterns it happens to have seen before.
A lot of people think much more concretely. I think there is a real value to be able to think about problems abstractly but it requires a breadth of knowledge, and you need to intentionally find new stuff to classify.