This version adds the new option to take verbose results from a standard run of qPTxM and use a random forest classifier trained on synthetic data to predict a (hopefully overlapping) set of modifications. The random forest we trained is distributed as a compressed binary -- decompress to a pickle to use it. We also include the scripts necessary to generate synthetic data, train and test this random forest so that developers can follow the same steps to train their own. Other changes of note: verbose output all_tested_ptms.out lists all modifications tested on all recognized nucleotides along with the relevant measurements (densities, cc, score, etc.) at each site.