Tools & Techniques Bioinformatician Have To Master?

Question

[Thaman]

Hi All,

We have constant discussion between our colleagues regarding the tools and techniques we have to know being Bioinformatician. It's been always hard to figure our what kind of scope does different tools available in market has got. In contrast, knowing all available Bio tools is out of question because there are thousands. Can BioStar people give their opinion regarding the tools and techniques that Bioinformatician have to master to account themself as a Bioinformatist in different research field?

Thanks

Answer 1

[Chris Miller]

I work on genomic datasets. I have colleagues who work on analyzing cryo-electron microscopy images. The overlap in the tools and techniques we use is minimal (almost non-existent). Really, the only thing we have in common is that we both use computers to analyze biological datasets, which is just about the definition of bioinformatician/computational biologist.

Trying to lump all of bioinformatics together is, I suspect, an exercise in futility.

[Dan D]

Yes, it seems like there's a trend towards "bioinformatics" being associated primarily with NextGen Sequencing. It's a broad category, and it's important to keep it that way to avoid confusion in the marketplace.

Answer 2

[Andra Waagmeester]

In my opinion does the answer depend on your specific research description. Personally I consider proficiency in Unix essential. I am still amazed on what you can do with only sed, awk, regex, sort, and uniq to just name some commands. I am a command line junkie, but again that is a personal choice.

For quick and dirty xml-parsing I use xmlstarlet (http://xmlstar.sourceforge.net/). With statistics, R is my friend.

For document and presentation preparation I use LaTeX.

Maybe more importantly then a set of tools and techniques is the ability to be sufficiently confident in any platform (that is Linux, Mac, Windows). A bioinformatician should be able to maneuver between the platforms other interacting disciplines use.

The ability to quickly learn is maybe more important then to learn a specific set of tools beforehand

[Thaman]

Command line is another powerful technique, thanks :)

Answer 3

[brentp]

1. Scripting in whatever langauge -- though a language with good libraries for bioinformatics is best. As a bioinformatician, you'll be dealing with large amounts of data, if you're limited to what existing command-line (or worse) GUI tools can provide you, you'll be unable to to anything novel. More specifically, a good skill is to be able to write script that can manipulate datasets that are larger than can fit in memory.

2. The ability to learn and utilize new tools. The field is changing and it is broad, if you're still using the same tools as 5 years ago--you're not going to keep up.

3. Some understanding/knowledge of statistics and biology.

4. linux

[Thaman]

Keeping update with changing technology is great point!

Answer 4

[Chris Evelo]

I agree with the general sentiment that every Bioinformatician is different and thus needs a different toolbox; plus the field develops at such a speed that you will want to use (and learn/develop) bew tools regularly.

For me the it is nevertheless important to have some idea since it determines what you would teach. And of course the teaching should be such that you become flexible and can indeed rapidly adapt to what is needed.

Some core things would be:

• a scripting language (probably Perl)
• a more structured language (probably Java, maybe Python to combine the two)
• R Bioconductor because there is uncertainty in the data and you will need some tools to not just handle the data but also the evaluate the statistics
• some data and database handling approaches
• a sequence alignment tool (BLAST, but it is more about the approach than the actual tool)
• some approaches to handle and evaluate 3 D structures (we never use those ourselves, but I think you should just know them, and they do produce nice pictures if nothing else)
• specific to what we do I would add pathway, GO and network analysis
• semantic web approaches (these are clearly becoming more important, so you should start teaching now)
• I would throw in a generic way to do identifier resolution (BridgeDB based), since that will save you loads of time in other things

I also agree that you just need to feel at home in the environment you work in so you need a good knowledge of the OS (Linux), the powerful basic tools available for that and a good editor and debugger plus a good mail tool since in the end you will spend a lot of time looking for and answering email.

[Thaman]

Now it's more descriptive thanks :)

[mylons]

I don't think you should automatically cast python into a more structured language. It IS more structured than perl, but it's capable of doing everything perl is without the overhead of Java or C++, but with the advantages of a strong OOP foundation and awesome built in class library. Perl is just way too obfuscated for long term use and maintenance, in my opinion.

Answer 5

[Laurent]

The most important tool for any bioinformatician is a good text editor. That's the tool you will use to write your code, inspect/debug your data file, interface with other tools (I'm thinking about ess here),... In addition to that, proficiency in the command line (which makes linux/unix a direct dependency, I suppose) is certainly an advantage.

[Thaman]

Thanks for pointing out about relevant text editor

Answer 6

[Lars Juhl Jensen]

I think this list is extremely short. There are many tools around, but I cannot think of many that one must know to be considered a bioinformatician. My list would have only two entries:

• A scripting language
• BLAST

I obviously do not consider everyone who knows a scripting language and BLAST a bioinformatician. However, the field is so diverse that everything else will depend on which topic you specialize in.

Answer 7

[lh3]

Even in a subfield, personal preferences vary a lot. I know quite a few Java and C++ programmers who happily dismiss scripting languages without problems. Once we learn Java, working in Linux is not an absolute requirement. Some of us know little about XML. Many of my friends and myself do not use R. A lot of my colleagues do not know a bit of latex. Sequence alignment is not indispensable for us who are not working on sequence analyses. The techniques we use are very different, but we can all be good at bioinformatics.

Personally I use C/Perl, work with command lines, write sequence alignment tools, love Latex and know SQL and a few other techniques, but none of them I think are essential also to others.

The only thing in common I think is that we all work on biological problems and that we all program - bio + informatics, which is how bioinformatics is defined.

[Thaman]

Good to know how varied jobs Bio field has got!

Answer 8

[Codemanky]

this PLOS Comp Bio article might be useful to find out the programming skills required for bioinformaticians. In addition to UNIX, scripting, and databases it also mentions importance of high performance computing and understanding hardware architectures.

[Thaman]

Thanks, but I have already gone through that article