Forum:What does a recruiter expect from a Bioinformatician ?

Question

[NB]

Hello,

I wanted to know what do PIs/recruiters/fellow bioinformaticians look for when hiring NGS-Bioinformaticians.

Do we expect the Bioinformatician to be skilled in wet-lab/sequencing and an expert in data processing with respect to NGS only or do we want someone who knows everything that is to know in Biology, statistics and coding and are able to carry out -omics analysis?

Speaking of coding, is knowing one language thoroughly good enough for now or do we also need to know to code C++, Java, R, MySQL etc? I know my colleagues when recruiting NGS Bioinformaticians expect candidates to know detailed coding which is of a developer's standard as well as develop algorithms. I feel these are more of computational biologists or developers than NGS Bioinformaticians.

Meanwhile, candidates who have higher education (Masters/PhD), good CVs and good publications in reputed journals on being asked on coding, say they know only shell for data manipulation and mostly string a bunch of open-source software together to generate reports (BWA-Picard-GATK-Annotation) and they need to look up the internet to write some basic codes - do we hire them?

It would be insightful to know what is expected from a Bioinformatician, specially when a few of us feel this

[GenoMax]

Are you asking this question from the view point of a recruiter or a candidate? Since you seem to refer to both sides alternately in this question.

BTW: It is perfectly ok in my book to be an applied bioinformatician. Someone who knows enough about data/tools/biology to make it all work, with minimal programming knowledge (and some help from Biostars!).

[NB]

a new recruiter... but could end up as a candidate at some point with time.

[Kevin Blighe]

Nandini, keep in mind that a job title such as 'NGS Bioinformatician' immediately implies that it's just an analyst role. You will most likely not have any input on the leadership or direction of the group. I see that job title used a lot in the UK, less so the USA (I'm also in London).

[NB]

yes I agree that a NGS Bioinformatician title in UK is very vague/general and that;s why my curiosity. We just conducted an interview last week and it was my first time being on the panel. The job was advertised for data processing and developing pipelines using open source software and validation - very basic. I was taken aback when one of us asked questions relating to a developer's profile

[Jean-Karim Heriche]

I don't think there's much of a general rule, it depends on the job, its requirements and its environment. Some PIs/recruiters have a clear view of what they need/want and some don't. Often recruiters also have to modulate their expectations based on the type of applications they get and their budget: highly skilled/competent/experienced people tend to be more expensive :)

[Kevin Blighe]

Yes, indeed, all I can say is that each role has a job spec and expectations of the candidate. If one accepts a role that's clearly advertised as a data processing analyst, then that it all that you'll be doing. Other roles may mention things like 'leadership', 'negotiating skills', etc, which would imply a more important and central role.

[steve]

They should be familiar with all the software and tools listed on their resume, and preferably be able to list examples of when and why they used each of them.

[Joe]

This is a good question actually. In particular the desire for bioinformaticians to have professional software development level skills is very offputting were I to be looking for a bioinformatics role.

I'm a wet lab biologist who enjoys learning to code and doing bioinformatics. I'm at the level I'd consider applying for a 'bioinformaticians' role, but I certainly wouldn't describe myself as a computational biologist or similar. I have no qualifications in anything CompSci or even in bioinformatics (excluding the fact that I'm doing some for my PhD).

I'd be interested to hear what group leaders here would have to say about this level of (in)competence!

Answer 1

[Istvan Albert]

I think first we must recognize that word "bioinformatics" may describe vastly different needs. The most important detail is the candidate's suitability for the actual requirements.

Even so, after having interviewed dozens of candidates and having followed the careers of many others I have discovered commonalities and indicators of potential. I like to do so called fizz-buzz tests as they apply to bioinformatics:

1. What is a FASTA file?
2. How would I access the last ten bases of a sequence stored in a FASTA file?
3. Write me a program that prints the numbers for the column 3 of a file in reverse order.
4. How many reads of a SAM file cover chromosome 1 at coordinate 100?

And so on. I found that asking easy questions is much more effective in helping me gauge a candidate's skills than asking hard questions.

Because all these questions only appear to be simple - there is a lot of depth to even the FASTA format - the better the candidate the more they know about FASTA, all of its idiosyncratic behavior, you can learn all kinds of fun details.

In a nutshell, people that can talk 20 minutes about all kinds of weird FASTA behaviors are good bioinformaticians.

[Anima Mundi]

Hi Albert, in my opinion with such tests you would disregard a bit the importance of a good biological background. As genomax says, my feeling is that a sound biological knowledge plus some computer skills can make a good combination (among others) for bioinformatics.

[Istvan Albert]

But that is not all that different from what I am saying, wouldn't you say?

I am not advocating that a bioinformatician should have the mad skills to write the next de-Bruijn type graph assembler - all I am saying is that they should be very comfortable reading a text file - in turn, they got to be good at that.

Not sure what you mean by "some computer skills". I will say that naive bioinformaticians: those that think that "simple looking" things are also trivial - now those people cause the most trouble.

[Anima Mundi]

Istvan, I totally agree on the fact that if you look close enough nothing is trivial, but that applies to everything. So if you select candidates on the base of their skills on apparently trivial topics, in my opinion you should at least give as much importance to "biological trivialities" as you seem to give to reading text files. By that token, I think that many computer scientists ending up in bioinformatics could (understandably!) have some difficulties with basic biological concepts (e.g. defining what a gene is). My feeling is that bioinformatics is one of those fields in which very few people can truly say to possess a complete background (and I am sure you would agree on that); therefore, maybe an especially patient and forgiving attitude is rather the way to go when judging each other's skills.

[Istvan Albert]

My take is focused on what comes first and what comes second. I want the candidate to understand biology as well.

But when someone applies to be a bioinformatician I want to know first and foremost if they have the ability/skill/experience to do the computational thinking.

There is a difference in learning biological concepts compared to learning the command line - the former is a more descriptive science - the latter is way a thinking. It is harder and slower to change the way one thinks than to understand then recall concepts.

Again I am not judging or claiming one to be more valuable than the other in general, or on the absolute scale. I'm looking at this from the point of view of hiring a bioinformatician. Bio may be the first word in "bioinformatician", but I think we can agree that informatics drives the process.

[Anima Mundi]

Just for the fun of discussing: what I see is that a recurrent theme among "hard" scientists (i.e. those having a background in physical - as opposed to natural - sciences) is that biology can be approximated to a list of notions. I disagree with that view: while notions in biology are certainly important, the low level of formalization should not be mistaken for lack (or paucity) of logical relationship between concepts. In my view, good biologists need to possess deep abstraction capabilities, to the point that one cannot easily quantify how much biology and informatics differ in terms of requirements for thinking over remembering.

As a biologist I regard the word "bioinformatics" as a... sort of pleonasm: life itself is ultimately a mere consequence of the rules of information flow. On more practical terms, still I feel it is subjective to tell whether bio- or -informatics is the driving force of a bioinformatics project, at least generally speaking.

It is likely that we could go on quite a bit with this kind of discussion, as probably this is one case in which radically different backgrounds meet ;).

[Joe]

I like this last sentence. That sums it up very nicely.

Answer 2

[h.mon]

Are you recruiting, or looking for a job - i.e. facing recruiters?

If you are recruiting, it seems you already know what bioinformaticians are and, in particular, you know bioinformaticians can't possibly know everything about bioinformatics. So you should know what are your requirements - the actual skills necessary for the problem you want to solve - and should try to find someone as close as possible to filling those requirements. Sometimes that means knowing wet-lab techniques and having a good background in biology, sometimes not. Sometimes it means knowing one programming language at depth, sometimes, being reasonably skilled at more than one language. And so on.

If you are a bioinformatician looking for a job, you will find everything, from nearly completely clueless recruiters to very knowledgable ones. I was asked once if I could code using the "samtools programming language".

[Kevin Blighe]

"samtools programming language".

Hey, I invented that ;)