.....Prior knowledge of a precursor ion’s charge state is propitious before submitting its MS2 to a search engine. If the charge state is unknown, the search must be executed for every charge state hypothesis and can generate a great computational overburden. To solve this problem, we present Charge Prediction Machine (CPM), a software for infering precursor charge state from ETD low resolution mass spectra. CPM has roots in the Bayesian decision theory and introduces methods to account for different co-fragmenting precursor ion species. It also carries a nifty graphic user interface as seen on the right; however, it can also be executed in the command prompt as to be integrated into bioinformatic pipelines (This version is available upon request).
The Windows CPM version can be downloaded at the downloads section of the Yates Lab website
CPM has recently been published in Analytical Chemistry
A PodCast is also available here.
Instructions
Just follow these steps....
1) place your low resolution MS2 in a single directory
2) select an output directory
3) choose a classification model. (available for download in the downloads
page as well)
Press Go! and thats it... then just wait a few minutes...
All the other tabs are unnesessary. The only reason to select the params tab
is in case you wish to use the classification model of +2 through +6.
If such is true, before pressing go, hit the params tab, click on "Load
default classification settings", and unselect the charge 7.
Thats it!
FAQ
I got 100 MS spectra and CPM outputs 150 Z lines; is this expected?
The gist of CPM lies in the relaxation parameter; it holds a compromise between accuracy, and computational cost. So for example, if a dataset has 100 spectra, and the relaxation parameter is set to 1.5, no matter what, the output will end up with 150 charge state hypotheses (150 Z Lines). What is really going on behind the scenes is that the relaxation parameter allows CPM to assign an extra 50 charge state hypotheses. CPM will optimize its resources to assign these extra 50 charge states to the ones that "it had most trouble" in an optimal way . Even if CPM correctly predicted the 100 spectra charges using its first hypothesis (impossible to know that for sure), it will "relax" as it was allowed to do so. The recomended relaxation parameter is 1.75.
