Hi everybody, I have HIV reverse transcription gene mutation samples .also I have mutation information and corresponding drug resistance of each sample. I want to train a machine-learning method in R to regress the resistance values for each drug on the viral mutations. but I don't know how to include the mutation position as features in my model . I want to use linear regression, SVM-regression, random-forest regression. My data looks like :
P39 P40 P41 P42 P43 P44 P45 P46 P47 P48
- - K - KN - R I - -
- - K - KN - R - - -
- - - - R - - - - -
- - - - - - - MI - -
- - RS - KN - - - IL -
- - - - - - - - - V
- - - - - - - MV - V
How can I encode this mutation information as value of my features in my model ? should I put 1 when there is mutation and 0 when there is no mutation in comapring with wild type sequence ? if so, the question is that, how can I incorporate the value of specific mutation in my model ? for example, when in the position p41 the mutation is "K" rather than "R" the virus show high drug resistance ?
Hi Azim. Since we know nothing about the the method/software that you are trying to use, it is very hard for anybody to give you suggestions concerning how to encode your mutation information. Please give us more details about your goal, what you already know, what software/approach you are using and any pertinent detail. This will help you get better replies.
Thanks Eric, I want to use R programming language . I want to try some regression methods like, SVM-regression, regression random forest, ....
Hi Azim. I deleted your 'answer' since you also posted it as a comment. Use the 'answer' zone at the bottom only to answer the original question and comments to discuss some answers or other comments.
The best thing for you to do is to EDIT your question and to add this new information there.