search

berlinguyinca / binbase

Automatically exported from code.google.com/p/binbase
0 stars 0 forks source link

required tool #61

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago 
Gert,

for compound ID (or cleaning up Bin spectra), we would need a tool that 
calculates the correlation matrix of all ions of a specific Bin within a set of 
specific chromatograms in which this Bin was found (i.e. not replaced). With 
that, one could see which ions really belong to that Bin, because the 
correlation should be above 0.9. 

So, basically all we'd need from your team is a tool that extracts the actual 
spectra from those set of given chromatograms and spits out those mz:intensity 
lists. Dmitry could then write an R-tool that would pick those up and gives the 
cross-correlation matrix. 
This is what John Meissen did with ion mz144 in his unknown that convinced me 
that mz144 was indeed part of the Bin

thx
Oliver

Original issue reported on code.google.com by berlinguyinca on 20 Nov 2013 at 9:43

GoogleCodeExporter commented 9 years ago 
basic query is:

select all massspecs where bin_id = ? and sample_id = ?

calculate a correlation matrix over them against a given bin spectra.

http://commons.apache.org/proper/commons-math/apidocs/org/apache/commons/math3/s
tat/correlation/PearsonsCorrelation.html

Original comment by berlinguyinca on 21 Nov 2013 at 1:11