How can a person with computational background join the longevity field?
People with computer science, physics, math and other computational backgrounds tend to ask the same questions about joining the longevity research field. As a person who have already done it (my own journey summarized here). Below, I’ve tried to summarize my experience and advice on how to transition to aging research as smooth as possible.
highlights of the transition from other computational fields to longevity research
- First of all, and from my experience from talking to different people, the most important reason driving them to join biology/longevity/aging is gained access to a more fullfilling and meaningful work, with the hope to be able to contribute to the betterment of humanity, its health and wellbeing. The pay in biology is usually significantly lower than in tech or pure AI applications, but getting the feeling that your life is not being wasted on sorting socks in a marketplace ranking, or showing people more catchy ads to buy those socks is priceless.
- Getting rid of golden handcuffs is also a huge boost to the morale of a new convert. Trading money and safety for an adventurous journey into biology and longevity is usually thrilling.
- Biologists are genuinely amazed by people who understand maths. Maybe, it’s a side effect of their fear of it, but anyway you can benefit from it. It’s the easiest way to make friends with biologists — help them with their data analysis, and run a few statistical tests for them.
- Your background and perspective would be unique and authentic, especially if you are the only computational person on your team. You may add a lot of value to general discussions.
- You may resque a lot of experiments by thoroughly analyzing their readouts. It happened to me quite a few times that experiments considered completely failed were rescued debugging data collection or preprocessing steps.
of course, there are also some potential lowlights which you may experience
- You’d be surprised how slow everything is. It’s not that they would think that they work slowly — no. It’s just there is a lot of grounding into real world. It’s more like cooking and baking. You may try to optimize your logistic, and ordering ingredients — you’ll anyway have to wait for an hour for a cake to bake. And you will have to follow a recipe. You can try to shortcut but most likely you’ll end up repeating everything from scratch and wasting even more money and time. It was extremely frustrating for me when I was starting but now I’ve learnt to be more patient, and multiply all my expectations by 5-10.
- Biological experiments are not super reproducible, and most of the times it is hard to draw any conclusive insights from them. Designing a good experiment with a clear yes or no outcome is a huge challenge. Even with a good design, the results may disappoint you
- The overall quality and volume of data available are limited. You may think that there is a lot of data out there, and thousands of experiments uploaded to various public repos:GEO, NCBI, etc. It is true, though technologies advance extremely fast, and most of ancient experiments (in this field, ancient is 3-5 years ago from now) will most likely end up in trash at some point of your data analysis due to relatively low quality.
- Batch effects across datasets due to multiple reasons — another measurement technique, another strain of mice or worms, different laboratory conditions, different vendor of reagents, different climate, chow, etc. Well, we study life here, it tends to adapt to different conditions, and have natural variability. Also, it’s alive and would respond to any changes in experimental conditions. It may sound trivial, but you’d be surprised how bad it is in comparison to most of other STEM fields.
- Biology is still not super quantitative, people working in the field would most likely have no interest or background in maths, programming, or anything you feel like all people around you understand. Using any quantitative concepts, equations, code would most likely scare any wet-lab scientist away. Of course, in almost any paper you’ll find plots, some tables and numbers, statistical tests, etc. Though, don’t get excited prematurely. Since you’re entering their field, it’s your obligation to adapt to their rules and learn to explain your thinking in a more human readable way without referring to technical concepts.
If you wonder what career paths or typical scenarios await you as a person coming from computational backgrounds to aging/longevity, I can think of a few.
Potential career paths for new converts from computations to biology/aging
- Standard bioinformatics, tool development, data processing pipelines. A lot of things have been already written about it. It’s basically software development for a very specific application. Be ready to learn a lot of obscure tools, languages, and pipelines. New sequencing, imaging and measurement techniques are coming out every year. There’s always a need for more analysis tools.
- Data engineering, infrastructure, IT/tech support in labs. Be able to debug code in R, help out with Excel and rebooting laptops. All kinds of support roles between a wetlab scientist and a computer.
- Data analysis as a service. Wetlab people ask you to process the data they produce, and provide some basic insights, plot a lot of graphs, present some tables, run some statistical tests. You may have varying degrees of freedom, but it’s a supportive role. You’ll be a convenient API for wetlab scientist to understand if their hypotheses churned out proven or not.
- Actual computational research. It’s neither IT/tech support, nor development of tools. You think about hypotheses (or A/B tests if it’s more familiar for you), find some data produced by someone else or beg wetlab scientists to produce more for you, test your hypotheses, improve and repeat. You can even try out wetlab yourself, if you find a person who can introduce you to this art.
Of course, it’s not black and white, and most likely you will have to do all those things to various degrees. For me it’s typically something like 10/10/80%. For others, it may be the opposite — 80/10/10 or 10/80/10, or any other combinations.