Forum:How Did You Learn Good Practice In Bioinformatics/Computational Biology?

Question

[epigene]

I came from a wet lab background and now doing a Ph.D. in genomics. Our lab is pretty decent in publication and my advisor often says the training we received in handling large amount of data is very desirable out there, etc. I can sense the need for this type of training is strong but I don't feel I'm getting the best or most effective training in how to handle data or being a good bioinformatics even though our program is ranked pretty high in US.

Just looking at my peers, I felt there is a great level of variation in the level of technical skills students have in the genomics filed. I struggle to find a "textbook" on the best practice type of thing. How did you learn it or suggest to do so in an efficient, systematic way, for students with biology, non-computer science background?

Answer 1

[Irsan]

Reading text books is a good start but the best way to learn good practice is by experience. Analyzing data is something different from reading about analyzing data. When you stumble on a technical difficulty that you find hard to tackle, in my opinion posting a question on a dedicated Q&A site like biostars or stack overflow gives you faster and better answers then asking someone in your lab/proximity (in my case at least, I'm the only bio-informatician within our section). Even when you have more general questions like you feel you are missing something or you want to hear about experiences and/or practices of others, Q&A sites (biostars in particular) help you a lot.

[epigene]

I agree, I was amazed by how quickly people respond on stackoverflow

Answer 2

[Medhat]

Hi,

As @Irsan said most of the time you will learn by doing that is the hard way, but it will give you a keen solid stand-in bioinformatics field, also to have a basic knowledge by reading books or watching lectures is good, somehow it paves the way and make things little bit easy to follow.

Here is the most useful rules that I think you shall follow;

Advice for newcomers to the bioinformatics field

Then please read this

10 Steps to Success in Bioinformatics

I want to learn bioinformatics! A guide for complete beginners.

A beginner's guide to bioinformatics - part I

A beginner's guide to bioinformatics - part II

and If you are alone or searching for new knowledge follow this;

A guide for the lonely bioinformatician

Lastly spend a lot of time in Biostars and seqanswers you will learn a lot, I hope that helps :)

[Eric Normandeau]

Thanks a lot. I got a mail exchange with someone who contacted me for advice about switching to bioinformatics. This will be handy!

[Medhat]

It is my pleasure you are welcome any time :) that is the purpose of the community

Answer 3

[GANI]

The following PLOS article summarizes many courses related to Bioinformatics in a systematic way. You can choose the courses based on you level of your experience in the particular field. I would suggest the Math and Computer Science departments for you.

http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002632

[epigene]

yeah, good to know this!

Answer 4

[Zev.Kronenberg]

Implement everything you can, within reason. Someone who can teach themselves new methods / algorithms is much more desirable than someone trained in "big data".

Also take as many statistics courses as humanly possible.

Answer 5

[lexnederbragt]

I can recommend these articles: Best Practices for Scientific Computing and So you want to be a computational biologist?. Beyond that, I would take a Software Carpentry Bootcamp, see http://software-carpentry.org/.

Answer 6

[epigene]

I realized that I asked this question 3 years ago. Now I've got a bit more experience and would like to share some of my personal thoughts on the question.

1. Learn by doing. I ended up picking up most of what I know by working on the problems and trying to find answers from the internet, mostly from stackoverflow and related sites, some from biostars. It's a painful and kind of slow process but I guess you have to go through it to really master certain skills.

2. Keep a good technical notes about all the problems you encountered. Either keep it to yourself or write it in a blog. Summarize them from time to time. I often revisit the same thing again and again later.

3. Read other people's code, especially the commonly used tools (samtools, bedtools, etc) and learn how they implement certain things.

4. Go to conferences. You will meet like-minded people, hear new development, tools in the field.

Answer 7

[Biojl]

Committing lots (LOTS) of errors and looking at other people's code.

Answer 8

[Csaba Kerepesi]

This is a very good and practical textbook with computer science basics: James Tisdall: Beginning Perl for Bioinformatics

[SES]

This is a great resource, but unfortunately that book is nearly 13 years old now. Here is a link to a free, modern perl book which is a bit more up-to-date.

[Medhat]

Sure this is very up to date book,But should not we advice with python better?

[SES]

That would be an opinion and not relevant for how to learn bioinformatics or be successful in the field. Without question though, learning to program from a book that is 13 years old will lead to a lot of problems, regardless of the language. That was my point, not that you should learn Perl. Sorry if that was not clear. You can be successful with any modern programming language and everyone can benefit by learning more than one.

[Medhat]

Totally agree sorry for miss understanding :)

[SES]

My comment was rather terse, it's my fault. I do think learning good programming habits is essential, like modern best practices in whatever your language.

Answer 9

[Sudeep]

This might also be a good read: How Not to Be a Bioinformatician

[Vivek]

Do not provide a graphical user interface: command line is always more effective. Force your end-users to use the command line

I don't understand this part, some of the most commonly used bioinformatic tools are command line based and do not come with a GUI.

Answer 10

[PoGibas]

You can/should try online courses (mentioned #1 in suggested PLOS article by GANI).
There is Data Analysis for Genomics coming up later this year. I already signed in and going to invest some time into this.

Take a note that prerequisites to this course are "programming skills, basic familiarity with the R programming language" and for this I strongly recommend learning sites like DataCamp and R-Fiddle.