Performance improvements.

[ypriverol]

Currently, msgf+ is used in combination with Comet by quantms pipeline for searching. However, in cloud infrastructure, it is desirable that the search engine runs faster and consume less memory resources. Any improvement we can make to the current version will impact quantms.

Some ideas from exiting search engines we can explore:

• (MOST Important)The fragment index store system implemented by MSFragger and other search engines. Here, you can see a nice explanation about the idea: https://www.nature.com/articles/s41467-019-11337-z.
• Related with the previous comment. When msgf+ index the fasta database for searching, do not included the ptms added to the search (Comet strategy) (see comment How to reduce MS-GF+ search time MSGFPlus/msgfplus#108 (comment))
• mzML loading: MSGF+ uses for reading the mzML files the library jmzReader. However, a new version of that library with faster reading implementation can be found here: https://github.com/bigbio/pgatk-io. Please benchmark before replacing the libraries.
• Other minor improvements can be about use more efficient java 8/9/11 features.

Other possible improvements:

• Training new models based on synthetic libraries from multiple instruments (see issue how to "properly" train new scoring model MSGFPlus/msgfplus#80)