How to enter the computational biology research field with a PhD in Neuroimaging?

Question

[hamaisa]

I am currently a third year PhD student in Neuroimaging. My project is divided evenly in the neurosciences and in medical imaging, specifically diffusion MRI, and is purely dry-lab. My work requires computational, programming, statistical and a bit of physics knowledge to effectively tackle research problems in this area. Before my PhD, my honours year involved drug discovery in the parasitology field. I changed from wet-lab to dry-lab after my Honours because I didn't like wet-lab research and was much happier crunching numbers on a computer.

I am very interested in making a switch to computational biology after my PhD because I love biological/clinical research problems and I would like to apply or develop my computational skills in the biology field, as opposed to imaging physics.

I have a few questions to ask from researchers/professionals in computational biology regarding this topic:

1. I understand that I will need to get up to speed with the computational skills needed in computational biology. I'm more interested in applying the skills to answer research/data problems, rather than developing software/algorithms without a research focus, so training wise, should I apply for formal qualifications after my PhD (i.e. a Masters in Biostatistics) or gain such knowledge through online training (i.e. Coursera and Rosalind)?
2. I am used to learning by doing the programming and statistics and teaching myself the computational skills needed for the problem. I do understand the argument that teaching yourself these computational skills is much more valuable and insightful than a University course, however, I would like to apply for computational biology Post Doc positions in the future. Given this aim, would a formal qualification be better in the long run than taking online courses (even with online certification)?

I've browsed this forum for topics similar to this one that I'm posting and I did find a topic with some great answers: (How to switch from bench research to computational biology?), however I'd like some feedback into my own personal situation.

Thank you!

[Zhilong Jia]

What skill sets are necessary to neuroimaging in general? Thank you.

[Zhilong Jia]

What skill sets are necessary to neuroimaging in general? Thank you.

[hamaisa]

In Neuroimaging, you need to be prepared to work on a Unix/Linux system and to learn Bash Shell Scripting and programming to run the commands necessary to run imaging software. With different imaging modalities (i.e. dMRI, fMRI, PET, MEG, EEG), you'll have different methods but the computing environment is the same if not similar. You should also have knowledge of neuroanatomy, which you'll gain if you run your analysis often enough, and you should have a basic idea of how the scripts for your imaging method work so that you can tweak them if they don't work as they should on your imaging data. So the basic computational principles of good data management and patience when programming/running and designing your pipeline apply here too.

[dariober]

I don't know anything about neuroimaging but from the way you describe it I would say you are already in the computational biology field! Maybe what you mean is how to enter the subfield of (next generation) sequence analysis, which is currently the main branch of computational biology/bioinformatics but it's by no means not the only one (others please correct me if wrong...).

My work requires computational, programming, statistical...

With these skills I would just follow Jean-Karim Heriche's advice: Just apply for a project you find interesting and be prepared to learn new topics.

Answer 1

[Jean-Karim Heriche]

Just do a postdoc in the field you like. Find a project you're interested in then select the best lab to carry it out.

[hamaisa]

Thanks for your feedback!

That was a surprising answer given that a number of colleagues I've asked say that computational biology needs preparation to get into from neuroimaging, and that I have no hope of doing a postdoc in that field if I don't brush up my skill set.

[Anima Mundi]

I think it is going to be hard, but yes, if you do not try you will never know ;)

Answer 2

[MAPK]

It will take you another 1-2 years to gain some expertise in the field, particularly in terms of programming/languages such as batch/shell scripting/linux/unix etc..but again it depends on how much you know already and how quickly can you learn them. Computational biology can be tough field with not so much satisfying career opportunities as you have to depend on someone (PI ) who brings funds and hires you in his lab. It's very rare that people with computational biology have their own lab and tap into funding sources.. careers in biological science in general is not so promising

[hamaisa]

My PhD work is all done using the Linux command interface and Bash Shell Script so based on what you said, I think I might be able to pick up. I'm currently teaching myself Python and I've used R for 3 years as well - do you believe a course in Bioinformatics would be useful as well?

[Jean-Karim Heriche]

Programming languages are just tools to get a job done. Since you seem already comfortable with the Linux environment and already use R, you'll not get much more on this front from courses. What you'll need to learn if you haven't already picked them up are data management skills but this is often not taught in courses. After that, a lot of what you need to know is really project-dependent and you'll build your skill set as you accumulate experience. You'll need to become familiar with the biology related to your project and the vocabulary that goes with it. A good foundation in maths/statistics is also definitely helpful in many situations. Keep in mind that for a postdoc you're not expected to know everything but to pick up the skills needed to complete your project. In this respect finding a lab with good mentoring/learning opportunities is a must.