Page 1 of 1

SVM training set format - how to convert 3D scanned objects (of protein scanning microscope) as proper SVM input

Posted: Thu Aug 30, 2018 11:54 am
by vinod
I use SkiKit-learn package to implement SVM and need to implement in Python following task. System input: 3 dimensional scanning of protein (density in each of coordinates of 256x256x256 pixels scanned sample). Desired output: 3D coordinates location of Amino Acids (which given protein comprised of, as its concatenated building blocks), and identifying Amino Acids type (from possible 22 types). Suggestion to use as training set: isolated scanning of known single Amino Acid from different angles, zooming, and resolutions - to get ("learn python" by SVM) its general geometric shape. And to so so with all 22 Amino Acids. And then, really analyzing of unknown protein with a provided chain of its Amino Acids sequence (as text), while demanding output as described above (AAs coordinates localization, and AAs type).

Please advise: 1. How to convert scanning samples of separated Amino Acids as input into training set? 2. How to classify output format (trivial is to numerate from 1 to 22 all groups and that's all? 3. Is SkiKit-learn package is proper environment to implement such a project?

Re: SVM training set format - how to convert 3D scanned objects (of protein scanning microscope) as proper SVM input

Posted: Sat Sep 08, 2018 3:43 pm
by velocity
I remember this from chemistry class. You should just use http://www.rcsb.org here you can view any protein and amino acids in 3D/2D and its protein chains. I can't remember but they probably have an API, if you want to use it in python