Best Python Support Vector Machine Implementation For Bioinformatics (Classifying Short Reads)?

Question

[Click downvote]

There are so many machine learning libraries for Python.

Which are good for classifying short reads with SVMs? It should include support for string kernels if I understand correctly.

It is hard for me to choose as I am new to ML. Furthermore it is difficult to compare/find all relevant options and info due to obscurity and lack of documentation.

Please advise.

[Sean Davis]

What is the biological question you are trying to answer?

[Steve Lianoglou]

Biology? We don't need no stinking biology ... we're talking machine learning, baby! Wooo hoooo!

[JC]

but if the question is just about machine learning, so it's offtopic, isn't?

[Manu Prestat]

It would be sad if one could not talk about string manipulation, alignment, classification, segmentation...etc in a bioinformatic forum. Don't you think? But you're right, whereas those are mandatory to analyse biological sequences, it's not a biology related question.

[JC]

I agree with you, actually I find interesting how he/she will use SVM for reads classification ...

[Manu Prestat]

Yeah, after reading your research interests in your profile, I was actually very surprised ;-)

[Click downvote]

It is a tool used by many bioinformaticians. Furthermore, I suspect I got a better answer here than I would have anywhere else. Whether it is ok or not is up to the mods.

[Click downvote]

"Is this a pirna cluster or not?" is the question I'll attempt to answer.

Answer 1

[Steve Lianoglou]

If it's string kernels you want, the shogun toolbox is for you.

Use the python_modular interface.

[Click downvote]

"also comes with a number of recent string kernels as e.g. the Locality Improved, Fischer, TOP, Spectrum, Weighted Degree Kernel (with shifts)"

Yes!