Pathogens are infectious microorganisms that lodge in a host and are responsible for causing diseases. In many cases, the detection of pathogens is expensive in resources and time. In this way, mass spectrometry is combined with data mining techniques to produce fast, efficient, and low-cost pathogen detection. An automated desktop platform named X-MassFP is proposed to analyze and train predictive machine learning models capable of identifying pathogens based on m/z data from mass spectra. Previous research analyzed serum samples from healthy and diseased salmonid fishes with Piscirickettsia salmonis. Their spectra were obtained and used them to perform a multiple alignment and binning experiments with our platform. Then, many combinations of pipes were implemented to obtain the best predictive models. Different bin sizes and feature selectors were implemented, as well as the use of oversampling on unbalanced data sets. The best results obtained with the X-MassFP platform corresponded to KNN using multiple alignment and SVM using the binning method, with 90% and 88.8% of accuracy, respectively.