Peptide Observation Model
EMSL Project ID
11492
Abstract
Washington State University (WSU) student Luning Wang will model the probability of peptides being identified in spectrometry as a part of PNNL's effort in increasing identification confidence in high throughput proteomics. She will do this research to fulfill her MS thesis requirement at WSU.The modeling effort will include the artificial neural network technology and statistics to predict the probability of specific peptides being discerned in PNNL’s liquid chromatography (LC) separations connected to mass spectrometry (MS) instruments.
The LC-MS and LC-MS/MS process has several steps that make certain peptides available and others not due to enzymatic digestion, ionization propensity, etc. These factors are not fully known, but can be estimated from the large peptide data bases at PNNL by examining what peptides are identified and which are not, in individual proteins that are assumed to be present in the samples. A protein is assumed to be present if a sufficient number of peptides from that protein are identified.
Mrs. Wang will research enzymatic cleavage rules, both theoretical and empirical, along with factors affecting ionization of peptides. Additional studies may include considering ordered/disordered regions of proteins to determine how multiple domains within a protein may affect what peptides can be discerned. These different sources of information will provide input to the prediction model, which will provide an estimate of a specific peptide being discernable at the end of the sample preparation, the separation, the spectrometry, and the use of identification software such as Sequest.
The model that Mrs. Wang will develop will be used with the discriminant function recently incorporated in determining confidence in peptide identifications. This discriminant function takes as input several parameters from Sequest and fuses those with a predicted value from an LC retention model developed at PNNL. Both the discriminant function and the LC retention model are in production at PNNL and provide confidence estimates to identified peptides. The probability model should increase the accuracy of the discriminant function.
Project Details
Project type
Exploratory Research
Start Date
2004-10-28
End Date
2005-10-13
Status
Closed
Released Data Link
Team
Principal Investigator