I'm not sure what it means that "the examples must be separated with tabs" and it seems like the sample (or gene) and its label should be separated with tabs.
I also think the word 'example' isn't specific enough and could mislead the user.
I'm also not sure what "gold standard status (matching labels)" means. 0 means novel prediction? 1 means predicted as positive sample, and -1 means predicted as negative sample?
Is the error function the penalty function for soft SVM? What is the tradeoff constant?
In terms of using SVMperfer, is it recommended that your input files are BIN files for performance? or is it optional?
Also, is can SVMperfer do nonliner classifications?