I’m a graduate student improving my skills in building computational pipelines for genomic data. I’d like to start freelancing to build a portfolio and gain experience working with real wet lab data, and have the intuition that anybody working with genomic data can make use of computational pipelines.
So far, I’ve reached out to a few PhD contacts in wet labs at other universities, but I’m wondering what other avenues might be effective for finding projects.
Do researchers advertise small-scale bioinformatics help within universities (e.g., through listservs or departmental boards)?
Have others had success offering services through general freelance platforms (UpWork, Fiverr) or domain-specific ones like KolabTree?
I’d appreciate any advice from people who’ve successfully connected with wet lab scientists for freelance or collaborative projects.
Ask yourself what your long-term goal is. We made the experience over the last decade that you can go two ways: You give data to someone who might technically be qualified but has no expert knowledge in the specific field or domain that you're working on, say a particular niche in immunity, or a specific cancer entity, and you will get lists and plots of results. Basically unreviewed or partially reviewed due to the lack of expert biological knowledge. This has limited value, as details often matter, and obtaining final publishable results that really drive a project forward requires constant discussion, redefining questions and adjusting analysis accordingly. For the projects we published, or are about to publish soon, there is no! way a freelancer could have contributed pretty much anything, other than do the preprocessing, such as aligning data, do basic QC etc. I need to agree that our field and what we do is very specialized, so it might not be representative, but we always get most meaningful results if we train the lead biologist to become at least a basic ChatGPT-assistet analyst, so they can hands-on see data, decide what makes sense and what doesn't and don't hit the wall finding a common language with an analyst from outside of the field. Some might argue that "blind" analysis from outside of the field comes with the benefit of being unbiased, but it suffers greatly from lack of expertise to know what does not make sense, and since biology is so vastly complex, the latter is the far greater risk ... as said, all in our experience, others might disagree, especially if analysis is very general, so you just do RNA-seq and all you need is a list of differential genes. Anyone can do this basically, AI-assistent if needed, today more than ever.
So what is your longterm plan? Do you want to become a pure service provider, or do you want to drive a project hands-on? If the latter, consider applying for a PhD, at best in a lab that allows to do both wetlab and computational work, as this is what the modern biologist needs.
That having said, what is your long-term goal?
The freelancing idea would be a way to get experience, while simultaneously providing other people with a service that they may need.
If I were to take the freelancing idea more seriously, it would be to give people tools to have more reproducible analyses. Rather than relying on a bunch of glued together scripts from 20 years ago, they have a modern analysis pipeline that they can rerun reliably and that will be standardized among other labs.
Of course I would need to make sure that there would be a good handoff with the lead biologist or whoever else will be using the pipeline. That means well documented code, thorough documentation, and maybe even a user manual.
This could be something done within the university that I am at, as I will have more credibility as a student at the same institution.
It's an illusion if you think that novel biology comes ready-made out of such pipelines. It's going back and forth with hands-on analysis. What can be automated is the preprocessing, but the actual downstream analysis, say starting from a count matrix, never will be. Preprocessing automation already exists, say nf-core or custom implementations. I use the same pipelines for preprocessing for years, it's always the same. But downstream never is.
It's always getting an idea from one analysis, validating, maybe re-doing, following up, and repeating. You might think it's different because of your, with respect, lack of real world experience, but there is a reason why basic research mainly operates on lots of custom scripts, because science and biology is so compex and noisy, therefore scrict automation in novel discoveries is so difficult, and often not remotely feasible.
Thanks for the info.