Careers in biotechnology, part II bioinformatics

Dear Madam,
I am very happy to see your very nice article.I am a student completed my PG Diploma in Bioinformatics and PG Diploma in Bioinformatics in addition to Graduation in Zoology.I have an Inetrnational certification in software testing also.if I have any opening in this pioneering Bioinformatics field for as a career option kindly advise me.Because I am very keen to know this.Anticipating your reply.

Sincerely yours

Abdul Jaleel.K

Abdul,

If I were you, I would be looking for positions in bioinformatics software companies. I think it would be very helpful to have people involved in testing software who understand the science.

Sandra

Fabulous post as usual and a great diagram. I do think you're missing one type of role. Back in the day, bioinformaticians in companies did everything from running BLAST to proving scientists with lists of compounds, but developments in software over the last few years has changed things a little. A number of companies employ informaticians whose role is methods development, either novel analysis tools that they can't get in off-the-shelf software, or implementing methods as part of workflow solutions. They work very closely with the software engineers/programmers and might very well be the same person in a different hat, but the role is different, IMO. Often these people tend to be algorithm developers, stats gurus or data geeks.

I personally think that the separation between scientists who use the programs and programmers who write them is extremely unhealthy. No one should be using an algorithm in their work they do not understand, from the fast Fourier transform to BLAST. If you couldn't write it, you shouldn't be using it.

The proper approach (if the program is teaching adequate mathematics) is to give a one semester course on programming where you actually implement some representative core of algorithms for the field. In physics this is some set of numerical methods for quadrature, differential equations, and curve fitting. What's needed is a book along the lines of Hamming's 'Numerical Methods for Scientists and Engineers' (by which he meant physical scientists). This assumes that biology students should have a decent grounding in analysis, probability, and statistics.

No one should be using an algorithm in their work they do not understand, from the fast Fourier transform to BLAST. If you couldn't write it, you shouldn't be using it.

That is simply absurd. Could you design a CPU? If not, I say you shouldn't be allowed to use a computer.

JSinger,

Fred does have it point, though it is a bit on the extreme side. It is very important for biologists to know what is happening to their sequence (or whatever) data besides input -> magic computer stuff -> results.

Understanding an algorithm and the conditions under which it is valid and how to improve its application by adjusting the relavent parameters is crucial. But that doesn't imply you could implement the algorithm in all its complexity in efficient, maintainable, testable code. Technology and science is a team effort. We don't all need to be able to do everything.

I have to agree with Jeb. I use and help develop new bioinformatic tools everyday (molecular biologist/biochemist by training) and yet I still need to be able to understand and explain why I'm using the algorithm/program, what it does and why the outuput is meaningful (assuming of course it is). I can't explain the code line by line but I can explain the overall process. Bioinformatic/computational tools are extremely helpful but you need to have at least a general understanding of what they do and how they do it (in my experience, this also means a decent understanding of statistics).

I think that the skills and educational background that are needed depend on the context where they will be used. It's ludicrous to think that everyone needs PhD-level training in mathematics or programming to do every kind of job. Nor do we need everyone to have the same level or kinds of training.

If you are doing cutting edge research or developing new tools or doing experiments, as Deepak describes, you need a very different skill set from say a data analyst who's screening samples for polymorphisms and reporting them to a supervisor.

There is some cutting edge research in bioinformatics going on in biotech (or at least in bioinformatics software companies), but it's not as common as the more mundane types of activities.

As Thomas points out, it's also important to note that in industry lots of work is done by teams. A team might include a biologist, a knowledge architect, people who are experts in data base design, people who can program, people who do web design, and IT people. No one expects everyone to have the same identical skill set.

I do think that if you're getting a PhD in biology and using bioinformatics, you should understand statistics and what the algorithms are doing. Otherwise you could end up misinterpreting your results. But I don't think you need to understand all the algorithms that you use.

But that wasn't really point of this post. My goal here is to identify the jobs that exist in the biotech world - outside of academics and see if my understanding of the jobs that exist is correct.

No one should be using an algorithm in their work they do not understand, from the fast Fourier transform to BLAST. If you couldn't write it, you shouldn't be using it.

Although I agree with other commenters that this is somewhat extreme (in particular, "understand" != "write"), I can sympathize quite a bit with it. I once found myself in the awkward position of being significantly closer to that viewpoint than the primary investigator in my lab was. This was NMR spectroscopy, not bioinformatics, so it was "magic radio frequency pulse stuff" rather than "magic computer stuff", but same idea. I wanted to know exactly what every pulse in the NMR pulse sequences did, at every point, since even a single error in the sequence meant no useable data. The PI, although very good at designing these NMR pulse sequences, didn't understand my need to understand. This was just one of many reasons I left this particular lab, almost screaming, a few months later...

Methods developers/programmers absolutely need to understand the problems they are addressing, and data consumers/producers also need an understanding of how they are getting their data (in an ideal world, they had input into the process), but the skill sets are different. How many biologists really want to understand the differences between two different machine learning techniques?

Also, sometimes its a question of time. I remember talking to someone at a pharma company a few years ago, who had published papers on some methods, and asked her why she wasn't using them at her current employer. Her answer was that she'd love to, but didn't have the time. However, if someone else did developed a workflow for her, she would happily be a user.

Thanks Deepak, it's good to have someone else commenting who can provide a view from industry.

I think you're always better off in a job if you have a good understanding of what you're doing.

The point remains, the sorts of thing that you need to know will depend on the job. I've found that academics and industry are very different places, so this is an attempt to help people who aren't in industry get some idea of the jobs that exist in the real world, and where people use bioinformatics in those jobs.

First an aside on the well-covered 'you need to be able to implement it to use it' reductio ad absurdum. The poster who commented that (s)he fled a lab where the PI wasn't interested in digging in as deep as the poster: that's the right thing to do. If you do enjoy that level of digging, that's great -- but it's silly to expect everyone to do that on everything.

Sandra's post & diagram are great. She asked me to comment, and I'll try to add a little bit.

The one key message I would try to work in is that some tinge of bioinformatics pervades most of biology today. Being at least exposed to the concepts will benefit a very wide range of career choices & give opportunities to shift direction in the future.

As far as people who would identify themselves as in bioinformatics, or would be identified by others as such, there is a very wide range of skills at the higher (doctoral levels). The rarest are folks who have strong bench skills and strong programming skills; they are a real find, but it is hard to juggle the two. Perhaps due to positive selection, another flavor are folks who are strong (but not expert) in programming, algorithmics and/or statistics but with no training or competence at the bench. This is the category I fall in: trained but not very good in the lab, very broad knowledge of biology, strong but not expert in programming & able to be reasonably competent in statistics (but no expert there).

But that sort of classification doesn't do justice to strong contributors in the field who fill other niches, or to the many niches which exist. For example, companies in the bioinformatics field have technical writers to write their manuals & trainers to train users -- both of whom benefit from actual experience with the field.

Corporate bioinformatics is going to run the whole range from one-person shows to large departments. In addition to what Sandra has identified above, I'd also point out that companies spend a lot of time keeping up on the competition, an activity which requires a lot of database searching and can be enhanced with better tools -- tools a someone with bioinformatics training might try to build (I once took a small stab as a favor for a friend). Another growing area is managing & mining data related to commercial activities, not just sales but things like adverse event data.

As far as training, there are some basics which every student should be exposed to. Being familiar with Excel is a must. Understanding some basic statistics is important, particularly such things as the tradeoffs of mean vs. median (funny, I was taught to calculate those somewhere in grade school -- but I think I was a grad student before anyone suggested why one might be better than the other for some situation!).

I think students should be exposed to at least the basic concepts of programming, but perhaps not actually have all students go through coding assignments. By this I mean that it is powerful to understand concepts such as linked lists, stacks, queues, etc even if you can't code them.

Database searching should be touched upon. A quick overview of SQL-style boolean/relational searching with contrast to fuzzy search systems (e.g. Entrez related articles, Amazon 'people who bought this...', etc).

Students would also benefit from being exposed to the idea of visual data mining, such as with Spotfire or Many Eyes (the latter has a free version; Spotfire is mucho dinero).

To some degree this is the spaghetti-against-the-wall approach to teaching -- if you expose a diverse group of students to a diverse range of material, most will key into something. Ideally one such course would be developed tilted towards biology students, another tilted towards Comp Sci/Stats/Math students, another towards Engineering students, etc. Computational Biology is too diverse & too rich to be monopolized by any specialty!

Wow! Thanks Keith for the fantastic and thoughtful comments. One of the main reasons that I started teaching bioinformatics in the first place was because I wanted my students to get proper instruction in using Excel.

(Before me, our "Scientific Computing" class was taught by this chemist/programmer guy who would do weird things like having the students convert all the values to logs for their y axes, rather than using the semi-log plots that we are all accustomed to using. )

Thank you for adding some kudos to the technical writers, as well. One of my next diagrams - probably next week - will cover the jobs that people do in bioinformatics software companies.

Hi,

Interesting discussion.

I have an undergraduate degree in business but I am in the process of a career change.

I have been admitted into a computer science degree program and for a minor I have selected a series of chemistry, biology and genetics courses.

Since I have already completed a 4 year undergraduate degree and will be starting a second undergraduate degree in a month which will last about 3 years, what do you think my job prospects in Canada would be like, alternately, which other places in the world are hotbeds for people like me.

Also, what kind of jobs would I be suitable for once I complete my education?

Any guidance would be greatly appreciated.

Thanks

What do you want to do? I would use that as a guide and look for places where you'll get a taste of that kind of career.

I don't live in Canada, but I think the biotech companies would have a similar structure. I think you should definitely plan on looking for internships.

Some good people might be Chris Hogue, Francis Oulette, or Micheal Smith. I think Chris Hogue runs some kind of bioinformatics boot camp. If I were you, I would try to attend.

Bioinformatics is an integral part of life science research. I guess nothing says more than that. I never had proper training in Bioinformatics, rather it comes with the job and I pick it up along the way. It works for me.

Nice post and great comments :)
Two things come to my mind when I hear the word 'Bioinformatics Specialist':
1. User
2. Creator
Of a Bionformatics tool(s).
The drive is totally personal on what one wants to be.
I get excited when I read about people like Prof. Eric Lander:
"earned a Ph.D. in math as a Rhodes scholar at Oxford. He was teaching economics at Harvard when he started reading about DNA. "Suddenly it was clear to me that all the beautiful complexity of life had simplicity at its core," he says. "This is the kind of thing mathematicians love"
[ http://www.time.com/time/magazine/article/0,9171,994017,00.html ]. Now he has huge team of mathematician, statisticians, programmers, physicians and biologist who are doing great work. They are specialist in their own domain and pick things up from other's domain on as and when required... I guess this is the model which works best.

You may be interested in this:

http://www.j-biomed-discovery.com/content/2/1/1

"Data management and integration are complicated and ongoing problems that will require commitment of resources and expertise from the various biological science communities. Primary components of successful cross-scale integration are smooth information management and migration from one context to another. We call for a broadening of the definition of bioinformatics and bioinformatics training to span biological disciplines and biological scales. Training programs are needed that educate a new kind of informatics professional, Biological Information Specialists, to work in collaboration with various discipline-specific research personnel. Biological Information Specialists are an extension of the informationist movement that began within library and information science (LIS) over 30 years ago as a professional position to fill a gap in clinical medicine. These professionals will help advance science by improving access to scientific information and by freeing scientists who are not interested in data management to concentrate on their science."

I work in an academic lab where we analyze genome-wide SNP data sets. The lab is entirely computational. My official title is programmer, but I think I'm essentially a bioinformatics technician. I spend most of my time manipulating data, implementing analysis pipelines, and creating/maintaining databases. If I were doing methods/algorithms development, I would probably call my self a computational biologist or computer scientist.

Keith's comment brings to mind this thought. A lot of us who became informaticians back in the day were not "trained" in the subject. I got my first job because I knew Fortran, Perl and had a decent knowledge of protein structure/function. The number of chemists turned bioinformaticians in my peer group was huge. Today, with more specialized, broader training, the demographic will change and to some extents labels will become less relevant, since people will have at least the basic grounding to perform multiple roles. That said, expertise in a particular area will never go away. There is just no time for someone to be excellent at everything even if you have the skill and/or desire.

Given the role of this blog, I'd definitely agree that biologists, bench or otherwise, need to know excel, some basic statistics, and have the ability to query databases, although I am not so sure about the latter as query interfaces get better and more powerful.

Nice article - and certainly it (and the discussion) capture some of the variation in bioinformatics.

I am a bioinformatics specialist from the in silico side (but of course my customers are all life scientists). But the niche my company fills is in mathematics and biostatoistical approaches to massively multivariate data. We don't consider ourselves to be software engineers (though we sertainly have to have that capability).

There is a whole other world of bionformatics which is not associated with databasing or data base integration, but which plays activekly at the interface between machine learning and modern statistics.

For me, its the most exciting area of bioinformatics - especially since it means I get to interact with some really cool new lab based technologies.

For most academic biology groups, however, being a bioinformatics specialist is a dead end job! People in these roles may or may not be PhDs, but they end up in fouth author hell - always the fourth author on hundreds of papers - whcih cuts no ice when it comes to institutional promotion boards.

I actively discourage young IT professionals and mathematicians from working in the area, noit because the challenges aren't there but because the career paths are too difficult.

Nice diagram--this pretty much matches the way things are at the large research institution I work at (as a bioinformatics programmer).

One comment: I think there's sometimes a tendency to view programming/software engineering as a trade skill like "advanced typing" or something. Many people who would be surprised if not aghast at seeing a programmer unschooled in biology running gels would think nothing of seeing a biologist cranking out code. In my opinion, the situations are not terribly different. I wouldn't discourage anyone who wanted to program from doing so, but I think it's important to realize that it takes quite a while to achieve a mastery of the subject.

Related to that, one of my significant tasks as a bioinformatics programmer--not mentioned in your diagram--is "trying to figure out and fix code written by scientists not trained in programming".

mkc:

Great points and you're absolutely right.

as far as:

Related to that, one of my significant tasks as a bioinformatics programmer--not mentioned in your diagram--is "trying to figure out and fix code written by scientists not trained in programming".

all I can say is RFLOL!

hi mam i am priya i have completed my graduatibuton in biotechnology and doing msc in bioinformatics.i read your article it is quite intresting,but in your article you haven't clear any where exact job profile for bioinformatics student.can you give the detail for this course and exact job profile for bioinformatics student.

Thank you for all the feedback, questions, and insights. I'm afraid that if I answered all the comments here, I don't think I'd have any time left for blogging. Look for more posts on these topics in the coming weeks.

/* Style Definitions */
table.MsoNormalTable
{mso-style-name:"Table Normal";
mso-tstyle-rowband-size:0;
mso-tstyle-colband-size:0;
mso-style-noshow:yes;
mso-style-priority:99;
mso-style-qformat:yes;
mso-style-parent:"";
mso-padding-alt:0in 5.4pt 0in 5.4pt;
mso-para-margin-top:0in;
mso-para-margin-right:0in;
mso-para-margin-bottom:10.0pt;
mso-para-margin-left:0in;
line-height:115%;
mso-pagination:widow-orphan;
font-size:11.0pt;
font-family:"Calibri","sans-serif";
mso-ascii-font-family:Calibri;
mso-ascii-theme-font:minor-latin;
mso-hansi-font-family:Calibri;
mso-hansi-theme-font:minor-latin;}

Just ran into this discussion on bioinformatics after coming
back from my vacation. That's why my post is one of the last. I work for
industry doing serious bioinformatics work and just want to comment on the
article and some of the comments made by my colleagues.

First, a few comments about the article itself and the
diagram attached to it. Sandra
brings out an interesting point by noting that "none of my biotech colleagues (with exception of one at Pasadena City
College) were ever programmers or IT people". Here lies a serious problem. For some odd reason they are not
referred to as fellow scientists (Computer Scientists), but allocated a role of
"code monkeys".
This is not just in this article, but all throughout the biotech
world. We, bench scientist, do real
stuff and they, computational people, are here with their "IT
plunger" because our "scientific toilet" overflows with
data. Sad, sad, sad!!! How about a
reverse situation where bench scientists are "lab monkeys" that
design and carry out experiments to support the discoveries made by
bioinformatics scientists using the data available to them or organized in a
different way?

The diagram is correct in describing a typical biotech
company. I worked for a few of them
and have to say that many of them don't really need a specially training
bioinformatics people. A few
"IT people" from green "Infrastructure" block would do
just fine for tasks outlined in red.
Companies that take Bioinformatics seriously create separate department
that mirrors "R&D" by its structure and role, one which exists
in parallel with Product Development R&D.

Deepak makes several excellent comments about the
state of the bioinformatics industry that I wholeheartedly agree with and
support. One remark on not having enough time to implement
algorithm is that it's a classic IT issue of project planning & management. Time is money and there's excess
of either better algorithms and concepts can be implemented.

Mervyn Thomas has
got it wrong. "Fourth author
hell" is purely due to the fact that people do support work and
don't attempt to discover something new and publish their results. You can't expect to receive a
Nobel Prize by pouring stuff into test tubes for someone else. You have to attempt and demand a bigger
task. That's how you get
promoted. If they don't give
it to you, find another position elsewhere.

On the topic how to educate a good bioinformatician I categorically have to say that (s)he has to have at least two bachelor degrees: one in life
science and another in computer science.
We don't have medical doctors with just a bachelor's degree
in this country. Why should we have
a fusion of two professions were corners are cut in education. No wonder then that they are stuck as fourth
authors for the rest of their careers.

Thanks S. for an interesting perspective.

I always thought the attitude you describe was more pervasive in academics than in the biotech industry. This discussion deserves a post in its own right.

If I make a distinction between software engineers & programmers and scientists, it's because most of the people who are writing code and "making things work" are not scientists.

I do not mean that in a derogatory sense. There is a lot of overlap, but I think calling all of the software engineers "scientists" would be like calling electrical engineers & electricians "physicists." They're not.

The problem is that some of the people who are scientists do treat people in the programming areas with a certain lack of respect for not acting like scientists. You said it yourself when you wrote that there's something less valuable about the people who

do support work and don't attempt to discover something new and publish their results.

The programmers and engineers, that I know, would see that as making a contribution to a team and doing a good job.

S.:
You mentioned my comment that "none of the people teaching in biotechnology programs were programmers or IT people - " and said that it was sad.

Well, there is a reason for that state and it's not sad, it's reality. Biotech programs provide training in lab techniques - cloning, PCR, DNA sequencing, using bioinformatics, protein purification, tissue culture, etc.

It simply wouldn't make sense to hire computer scientists or programmers to teach people how to run gels or operate fermentors.

I do not mean that in a derogatory sense. There is a lot of overlap, but I think calling all of the software engineers "scientists" would be like calling electrical engineers & electricians "physicists." They're not.

Khmm, khmm!!! I disagree! "Software Engineering" is a science. "Electrical Engineering" is a science. People get Masters and Doctorates in these disciplines. If they discover and publish something new in scientific journals they are scientists. Not all "physicists" are scientist either, just because they got a degree in Physics.
===========================
On a completely different note, I can't think of a reason why bioinformaticians would run gels and operate fermenters. May be we had a different program in mind, I don't know. I was referring to bioinformatics education, and it seems that you were referring to biotech education in general.
===========================
"Team work" goes a long way but unless it's well rewarded (such as regular IT work), the only reward in bioinformatics is purely scientific. You have to make an effort to make a name for yourself through publications, patents, and discoveries.

S: How do you define who's a scientist and who's not?

I would define it as a frame of mind and by whether or not you use the scientific method and approach what you do from a scientific mindset - that is: do you include controls? do you think about how you're testing your ideas? can you reproduce what you find? can you describe what you did well enough so that others can reproduce it? do you seek to understand the underlying principles behind what you found?

I agree there's lots in common between doing science and doing software engineering, and certainly, software testing, but they are not necessarily the same thing.

Hello

What are the openings available for a Ph.D in Computational Biology in Canada?

Are academic institutions the only option?

Thanks

Srini

That's why they call it BioInformatics. Combination of Biological sciences with Computer Science. The algorithm part of the science is very crucial for the kind of skill required for BioInformatics. Of course you do not have to understand algorithm to be able to work with it depending on what level of work you have to do with it. for e.g.. writing my comment on this website and for it to be shown involves a complex algorithm that many people do not understand and will never understand; but does it mean that they won't leave comment on this site no...It simply means that they do not understand what goes on behind the scene that is allowing them to leave a comment on this site. Does their lack of knowledge on how it works makes leaving a comment on this site irrelevant of course no. That's why they need people that are in the computer science field who understands how the various algorithms work to be key part of the BioInformatics field other wise the whole idea will be useless as many Biological science and related field do not have a clue on how database works and will be totally lost when it comes to actual design, programming and creating the various relationships when dealing with databases.

Software Engineering/Computer science is a highly complex field that requires a lot of attention, concentration and in dept studying for one to be able to relay to the principles behind database and database design. You may desire to be a BioInformatics etc but if you are not ready to handle the level of work involved to achieve that academic goal then you may have to consider another field that is less challenging cos there is no high level computer course work that is not challenging.

Software Engineers and Computer science graduates with Masters or Ph.d degrees are trained scientist but where they work determines how you will address them as scientist or engineers. The only problem though is that graduates in the computer science or software engineering field do not want addressed or see themselves as scientist even when they work in scientific research fields. Graduates in the above field generally have a low key kind of behavior. Status and big names do not mean much to them. So biological science people can boast all they want about being scientist etc...Computer science and software engineers could care less. That's my observation.