We provide 5 pre-trained model:

LC-ESI-QTOF-CID (1027 spectra) 
LC-ESI-ITFT-CID (447 spectra) 
LC-ESI-ITFT-HCD (2655 spectra)
LC-APCI-ITFT-CID (295 spectra)
LC-APCI-ITFT-HCD (882 spectra)

To use the models, go inside the directory and put your test spectra into the 
"test_data" foler and then "python test.py", the "result" folder contains 
the result.

Read the following before usage:

0. The libsvm 3.17 python interface is required. 
   https://www.csie.ntu.edu.tw/%7Ecjlin/libsvm/oldfiles/
1. The test data should be in MassBank spectra format.
2. The result file may contain None if the compounds is not in the database. 
3. We used a 2010 version of Kegg DB, where you could find all the Kegg  
   compound id in the "examples/MODELDIR/util/kegg_mass".
3. If you want to replace the DB with your own, you should replace kegg_mass 
   and kegg_fp.dict. The format of these two files can be found at 
   http://research.ics.aalto.fi/kepaco/fingerid/about.html#DB%20extension
4. If you want to replace the type of fingerprints, you should replace the 
   fingerprints DB in kegg_fp.dict as well as the train_output.txt for each
   model.