Hello kind bioconductor helpers,

I am fairly new to both machine learning and rnaSeq analyses so please excuse my naiveties. I am trying to design a classifier using a random forest. At the moment i train my classifier on three class (A,B,C) which is working well. I now wish to make a classifier which further splits class A into two subclasses (A1, A2). I was wondering if there is a standard way to do this kind of hierarchical classification (i.e. first split into class A,B,C then to split to A1,A2). I have been using simply the randomForest package in r, but was considering using the MLSeq package in bioconductor. I have potentially two ideas of how i could do this from my basic understanding:

1) Simply build a classifier for A1,A2, B, C (i.e 4 classes, and assume that anything falling into either class A1 or A2 would have been classified as A had i just trained on A,B,C classes.

2) Build two classifiers using a randomForest. I.e. first classifier classifies into A,B,C. Then i would build a seperate classifier to split the data falling into class A into A1,A2

Hope that makes sense and i look forward to hearing the communities advice and suggestions before i start trying to reinvent the wheel or do anything silly/unecessary.