MASCOT, everybody using it... well, almost. While looking for information about mascot, I found out that there is little available on the internet. In fact, I had made a preview page of me wanting to make a lecture about the mascot algorithm which did show up shortly later at "google" it rank 12. That induces the notion that there is not much available as web information about the mascot algorithm

Looking closer we see first references to MOWSE as early as 1993 and the techniques were in use even earlier. The MASCOT algorithm is mentioned only as early as 1999 - and the first books about MS algorithms dating to 2001. That makes it so the lectures in the year 2003 are relativly new on the subject. The knowledge about the details had not been covered by public lectures beforehand. Just some books take notice.

So what makes the MASCOT algorithm so successful that everybody wants to use it in 2003? Firsthand, everybody knows the MOWSE algorithm which puts a score on the protein MS fingerprint. The basic MOWSE algorithm is relatively easy to get to, it can be taught in a single lecture. And that includes hints as to where MOWSE may go wrong with sytematic errors. It had always been the question whether it would be possible to counter the systematic errors of MOWSE scores about MS fingerprints.

The MASCOT algorithm is based on MOWSE score but additionally to the results it display an absolute probability result that provides more confidence into the result of the experiment and computations - or makes the user to rework the experiment and/or computation parameters. (or so I read from the "success" materials spread around the net). Moreover the MOWSE results contains hints as to chemical neighbours, i.e. those with amino-acid losses as probable by the digest method. That makes the next probable items in the search slightly different than the mere distant check according to the MOWSE algorithm.

Now what, how does the training work?

• links
• http://www.matrixscience.com/help/history.html
  the link that states that MASCOT-I is actually the same as MOWSE-III !!
• http://www.matrixscience.com/pdf/Brochure_01-2-2002.pdf
  an overview restating what I found out in some other places as well.
• http://www.informatik.hu-berlin.de/Forschung_Lehre/wbi/research/ stud_arbeiten/finished/2003/proteomfactory_arbeit.pdf
  ProFound - Studienarbeit - weitgehende Beschreibung
• http://www.cgen.com/science/Publications/AnalyticalChemistry020103.pdf
  Extension on Probabilistic Scoring - including MS Intensity Values.
• http://www.hh.se/staff/bioinf/publications/Handout_PiumsPoster.pdf
  And another one that claims to be better than Mascot, trying to prove it.
• http://www.oardc.ohio-state.edu/tomato/Proteomics%20Presentation.pdf
  Some different pictures and diagrams for an introduction, nice...