[FreeKill]

If you really want to get a feel for how deluted the Bioinformatics community is, look for a job in the field as an outsider. It's not uncommon to see requirements like:

"Must be an expert in 18 technologies" "Must have a PHD in Computer Science or Molecular Biology" "Must have 12 years experience and post doctoral training" "Pay: $30,000"

It's delusional because they apply the requirements it took for themselves to get a job in Molecular Biology (long PHD, post doc, very low pay for first jobs) and just apply it carte blanche to all fields that may be able to aid in their pursuits. Especially when it comes to software engineering where it can often be extremely difficult to explain why you did not pursue a PHD.

[tom_b]

As someone on a bioinformatics team in a public research institution, salaries range from $75k to $100K for developers on our team. This includes a number of people, including myself, who do primarily normal IT things (data management, small webapps for various clinical and research needs) and also for devs doing pipeline/workflow mgmt software, novel dev (e.g. new research code for sequencing), and variant calling work.

In my geographic area, this salary range is somewhat below corporate IT work (say 10% to 15%), but generally higher than the typical university software dev job listing. The university is really bad to list jobs and job requirements with laughable salaries. I have seen (in other departments) web app dev jobs that require significant front-end and back-end skillsets/experience and then pop a salary that is full 50% less than entry level jobs for CS undergrads.

One problem is that hiring departments in that position will find someone to hire at that rate, so they think it was correct. From personal experience, I can verify that "good on-paper" candidates with exceptional credentials (say MS in CS, bunch of experience) from other depts who look to join our team are unable to to write any code at the whiteboard at all (say a for loop in java to println something). But to be fair, a recent job interview cycle one of my teammates performed produced exactly two candidates out of 16 who could do this and only one of those could write a SQL statement that required a simple inner-join. Most of those folks were external, so it's not just a problem inside the institution.

I have a number of cynical and embarrassing opinions about this situation.

[tensor]

I don't disagree with most of your post, though I cannot resist commenting on the whiteboard interview test. Writing anything other than rough pseudo code or algorithm sketches on the whiteboard is a silly exercise. It's not reflective of any sort of reality, probably indicates to candidates that you are not working on any interesting problems, and people won't remember exact syntax or library functions for any language that they don't use fairly regularly.

The whiteboard is only useful as an aid in explaining an algorithm. If a candidate can do that without the whiteboard, even better.

[tom_b]

I'm kind of with you on the whiteboard code issue (I was sitting in on the interview in question), especially for a "hard" coding exercise.

My bigger concern is that for a job that specifically highlighted the need for at least some SQL skills and some Java expertise, a candidate that can not, even after prompting, write a for loop in Java (or in any language, when offered the chance to do so in a "favorite" language") or write a SQL statement that joins two tables probably can't do much of anything, let alone work on interesting problems.

Here is the cold, hard truth - I know, both because of my own limitations and the opportunity of the job, that we are not going to get top % hackers. But if you apply to a job where the primary need is coding in blub, I think its fair to expect a simple question or two about basic blub constructs. I myself would be nervous about whiteboard coding for something complex, but also generally offer (in a cover letter) to provide some code examples to talk through at an interview ahead of time.

I think it behooves us all to have at least some baseline expectation to demonstrate some competence. Remember, I'm not thinking that whiteboard coding of an algorithms or anything.

I think a very fair (and concerning to me) insight might be: if you can use Google and an IDE, can you do all that this job requires?

[michaelhoffman]

I assume these are separate requirements. I have not seen any doctoral-level positions advertised for a salary of $30,000. The minimum NIH salary for postdoctoral trainees is more than that.

It's only delusional if they can't find people to fill the jobs. The idea that, as an outsider, you know what requirements they should use in their hiring process better than they do is perhaps more delusional.

[FreeKill]

I'm not an outsider and the 30K was a bit of an exaggeration, and I apologize for that. The point I was trying to make was that if you look in as an outsider, you would see the requirements being extremely daunting compared to what you might see elsewhere with a pay scale that is very low and unappealing to anyone who might match it. Unless, of course, you just finished your degree in some biological discipline where the jobs are scarce. They are absolutely delusional (and so am I, most likely) because in most cases what they really need to solve the problems they have, is the same type of person most companies would need in a similar situation, a quality software engineer with experience building quality applications that are both extensible and maintainable.

I worked in bioinformatics for more than 10 years before I moved on, and In my experience they do have a lot of trouble finding people to fill positions, especially outside of massive government funded groups like the NIH. This often results in passing on competent software engineers with a B.Sc. that don't meet the requirements in favor of PHD level biology graduates who have taken a year or so of undergrad computer science courses. In my experience, this leads to many of the problems discussed (and exaggerated) by the OP. While some of these people are smart and produce good work, much of the time they produce poor quality software that gets the job done, but as inefficiently as possible and they leave a code base that is virtually unusable. Overall, I mostly just wanted say that it's a mindset they REALLY need to get past for the long term success of the industry.

[kevin_rubyhouse]

If 30k is the inaccurate number, what's the accurate one? I'm curious as to what the realistic requirements are from your experience with the field.

[FreeKill]

I've seen a lot of job listings, at very large companies and academics for the 45-50 range. Keep in mind, these are jobs requiring a PHD, 10 years of experience, and a dozen or so technologies.

It's not really the money that's skewed, it's their idea about the person they need for the job. They don't need someone with that background (most of the time), they just need a junior level software engineer in which case the pay scale may not be too bad. There's a problem in realizing this, however, when the standards for your own field (molecular biology for example) are extremely high, so you expect it of all others as well...

[puredemo]

50k-60k starting out.

[michaelhoffman]

The field of bioinformatics will be fine even if there aren't any changes. We'll just continue to muddle through as we have. I'll agree that things would be much better if software quality were to improve, but changing that will require a change in incentives. Namely, journals or funding agencies will have to start requiring quality software.

[leont]

Check the Sanger Institute's job page (https://jobs.sanger.ac.uk/wd/plsql/wd_portal.show_page?p_web...). They offer «£29,750 to £37,525» for a "senior bioinformatician", for example.

[Devilboy]

He's exaggerating about the 30k of course but it's true that these positions don't pay very well compared to what experienced programmers can get elsewhere.

[Lost_BiomedE]

I have seen this in many bio fields. As the biology research becomes more of a computational problem, requiring unique solutions for bleeding edge research, I imagine the field is going to have huge pains before actually paying for the work vs. using seniority and degree level as the sole determinate of pay scale.