Advice - from wet lab to bioinformtics

[krespim]

I need some career advice and maybe you can add to some talks that I already had with my supervisor and some bioinformaticians.

My 1st postdoc will end in June and and I'm working in regulation of gene expression using a combination of custom arrays, RNA-seq and ChIP-seq and although my background is as a wet lab biologist, working in the computer (scripting, data analysis) has been a lot of fun. So am thinking about developing my bioinformatics skills to become more of a bioinformatician.(Note: Since the 1st part of the project went to waste, It is unlikely that the project will yield a publication by the time I leave the lab)

At the moment I do most analysis in R and shell scripting. My plan is to use the rest of my time and funding (I've some funds for career development) to learn more about Next-gen sequencing analysis while working in my project but also take some courses on the subject (I've already done one at the EBI) and learn another scripting language (I've started with python on my "free" time).

The questions:
1. Assuming I learn python (or perl in the next months) would that be enough to apply for pure bionformatics jobs (analysis of biological data)? I know it is unlikely but I still would like to hear your opinions.

2. In which courses, scripting languages should I focus? The biological questions I am interested in a cross-regulation of transcription and pre-mRNA splicing, namely in the context of tissue differentiation or response to stimuli, and RNA splicing-driven evolution. I'll probably attend a statistics course early next year and had biostatistics at Uni. I've learned TurboPascal but that was years ago.

3. Related to the above, could you recommend some courses?

4. I've suggested to take a break and do a proper bioinformatics masters. Would that help? I can see that focusing only on learning a new set of skills does has its advantages, but I prefer in-job training. Are you aware of Post-doctoral programmes for Biologists moving into the dark arts?

One sensible thing that someone told me was that I would find it very hard to compete with hardcore computational biologists, but I had advantage in that I could carry a project from start to finish, that is, make predictions, test at the bench (or design the experiments to) and then go back and some some analysis. My supervisor also mentioned that one asset that I have is being "bilingual" in meeting with bioinformaticians Are those valued assets in academia/Industry?

Thanks in advance for your input.

[adaptivegenome]

I really think you can succeed without further schooling. If you are concerned about getting a job, look for jobs you would want and then see what the requirements are. What degree do they want and what skills? Then be sure to learn those skills.

[FrieMueh]

hi,
I'm going down a similar road - studied molecular biology and ended up doing my PhD in bioinformatics now. If you enjoy scripting and data analysis more than actual wet lab work, then definitely keep on doing it! People usually turn out to be quite productive when they're doing things they actually enjoy. In my case I feel much, much more useful in front of the computer than standing at the bench eventhough I was trained very well for the latter alternative.
In our bioinformatics department, we mainly use python, perl, R and shell scripts - which language is the preferred one totally depends on the person.
Also, we're definitely looking for people who are eager to work in very close collaboration with biologists, so it's is absolutely imperative that the bioinformatician speaks the same language as the biologists - which, very often, is only the case with people who used to be biologists... Therefore, I absolutely agree with the previous post: just be confident, look for positions you think sound like the kind of stuff you'd like to do and then get in touch with the people offering the positions and sort out what they want and what you can offer. For example, this one here at our Institute: https://haufebewerbermanagement-2444.../Description/1
Cheers,

Friederike

[chadn737]

I've also been on the same path (still on that path). I think it really depends on how far you want to take it and where your interests really lie.

I think the key question to ask is whether you are more interested in the informatics/programming or Biological questions where you can apply informatics/programming. The former is less about the Biology and more about the development of tools and algorithms to be used by Biologists, the latter is still focused on the Biology but means a knowledge of the tools/techniques, their application, and how to work with those individuals dedicated to tool development.

If you find yourself more drawn to the actual informatics/programming, then I think I would go for a Masters and as you will need to know far more than scripting and also understand different algorithms and how to implement them. In my PhD I took several classes on algorithms in Bioinformatics...it was intensive to say the least and requires a very different mindset than Biology.

If you are like me, then you are less interested in developing some new algorithm and primarily interested in the Biology, using bioinformatic tools. This of course requires some programming knowledge to really handle the data and even write some less sophisticated tools to do analysis. In this case, I think an understanding of Statistics is more fundamental than anything, because often what you are attempting to do is some large scale Statistical analysis that can give Biological insight. If I had to do everything again I would have tried to minor in Statistics either as an undergrad or in my PhD as my school had a graduate minor in that.

On what things to learn. R is a must, especially for most of the statistical analysis. I started out with Perl and I know it has a lot of advocates and I will probably get attacked for saying this, but Perl is messy and reading other peoples code is nearly impossible. Recently I have started learning Python, and found it to be clean and logically simple for someone primarily concerned with Biology.

Either way, here are two great sources, one for learning Perl and (Unix), the other is an online course in Python that does an amazing job of teaching the very basics of Python and its structure:

http://korflab.ucdavis.edu/Unix_and_Perl/
http://www.codecademy.com/tracks/python

I would really recommend playing with the second link and start learning Python.

I can't tell you what is enough for Bioinformatics only jobs. I'm split between the computer and bench about 60/40, so there is still a need for me to be doing lab work.

[krespim]

Therefore, I absolutely agree with the previous post: just be confident, look for positions you think sound like the kind of stuff you'd like to do and then get in touch with the people offering the positions and sort out what they want and what you can offer. For example, this one here at our Institute: https://haufebewerbermanagement-2444.../Description/1
Cheers,

Friederike

Thanks for the feedback. Actually, I had already looked into that position and I am considering to make an application.

Maybe we'll meet if I go to an interview there

[krespim]

I've also been on the same path (still on that path). I think it really depends on how far you want to take it and where your interests really lie.
(..)

On what things to learn. R is a must, especially for most of the statistical analysis. I started out with Perl and I know it has a lot of advocates and I will probably get attacked for saying this, but Perl is messy and reading other peoples code is nearly impossible. Recently I have started learning Python, and found it to be clean and logically simple for someone primarily concerned with Biology.

Either way, here are two great sources, one for learning Perl and (Unix), the other is an online course in Python that does an amazing job of teaching the very basics of Python and its structure:

http://korflab.ucdavis.edu/Unix_and_Perl/
http://www.codecademy.com/tracks/python

I would really recommend playing with the second link and start learning Python.

I can't tell you what is enough for Bioinformatics only jobs. I'm split between the computer and bench about 60/40, so there is still a need for me to be doing lab work.

Glad to see that I am not the only one out there and great point you made about knowing what one wants. Like you I am also in the camp of "primarily interested in the Biology, using bioinformatic tools" and that 60/40 sounds about right for what I want - probably increasing the time scripting with time/knowledge.

Luckily my R skill are already ok. Shell and Python I've been learning with the book "Practical computing for Biologists" (which is excellent btw) but I'll definitely look into the links you've sent me.

Cheers.

[SrCardgage]

I'm writing this as a programmer, so keep that in mind.

First off, best of luck and congratulations on finding your passion! It took me awhile, too.

After learning a scripting language, it can seem remarkable easy to "hack together" a script that does what you want. And sometimes, this is all you need and that is good.

But if you are considering spending a large amount of time writing code, especially if dealing with large amounts of data, please spend the time to take some "hardcore" programming courses. What I mean are things like "Data Structures", "Algorithms" and possibly "Operating Systems Fundamentals". (these are not exact course titles, but just give you an idea)

I say this because I've seen the absolute mess people can make when they just continue to hack together code, without any coherent, driving design to keep it together. It is a pitiful mess that will make any sane person cry.

When you take these more in-depth classes, you will begin to learn how to think like a well-behaved programmer. In the beginning, it will seem to take FOREVER to design a program/project, when you could just hack it together, But weeks and months down the road, you will be able to modify your code MUCH more easily when you have a coherent design. You will also often find code you had designed for one project will be easily used in another. Now, that is true time savings!

And, yes, you can learn these things from a book, but IMO, it's better to take a class from someone who's been there before.

Cheers!

[GenoMax]

I say this because I've seen the absolute mess people can make when they just continue to hack together code, without any coherent, driving design to keep it together. It is a pitiful mess that will make any sane person cry.

Cheers!

That is actually by design: http://www.scfbm.org/content/7/1/3

[krespim]

That is actually by design: http://www.scfbm.org/content/7/1/3

That is brilliantly true

[krespim]

Thank you for the feedback.

I say this because I've seen the absolute mess people can make when they just continue to hack together code, without any coherent, driving design to keep it together. It is a pitiful mess that will make any sane person cry.

Cheers!

Actually I've already experienced one tool that only works in the developer's computer (true story); and another that although very good, it is written in perl and it has several separate scripts that need to be run before getting the results. Every time I use it I can't stop thinking "if should write a wrapper with at least some interface". Maybe I will soon.

[FrieMueh]

I am quite certain that we would meet if you were to come to an interview : )