Grant: Master's Award: Canada Graduate Scholarships
University of Ottawa
$17,500.00
A novel machine learning approach for real-time determination of peptide identifiability in mass spectrometry-based proteomics experiments
FY 2023-24
Summary
Proteins have diverse functions that shape cell behaviour, making their identification of great interest to understand cellular processes. Large-scale protein identification is commonly performed through mass spectrometry-based proteomics. Prior to mass spectrometry analysis, proteins are digested into peptides to facilitate their identification. These peptides are then fragmented by the mass spectrometer, which acquires a mass spectrum of the fragments from which a software tool can identify the peptide amino acid sequence. Due to the large number of peptides in complex biological samples, mass spectrometers do not have the time to collect mass spectra for all peptides, leaving many peptides and their parent proteins unidentified. Additionally, mass spectrometers often acquire mass spectra for peptides that the analysis software will be unable to identify after the completion of the experiment. This proves a waste of resources that could otherwise be used to analyze identifiable peptides and allow the characterization of more proteins. I hypothesize that an algorithmic approach can be developed to improve mass spectrometry’s identification capabilities by predicting in real-time during an experiment whether a given peptide will be identified if the instrument collects its mass spectrum. I will design a supervised machine learning algorithm that will use peptide properties collected by the instrument prior to fragmentation to predict, in real-time during mass spectrometry analysis, which peptides are most likely to be identified if a mass spectrum is acquired for them. My model will optimize mass spectrometry resource usage to improve protein characterization in biological samples. It will yield a better understanding of the active biological processes in such samples.
Department
Program
Master's Award: Canada Graduate Scholarships
Recipient
Location
Ontario, Canada
Recipient Type
Academia
Type
Grant
Source
CIHR Funding Decisions Database
As published
Id
506014
Name
El-Azouni Maram A
Region
Ontario
Country
Canada
Orgtype
Institution Paid; Research Institution
Prcname2
Special Cases - Awards Programs
Projecttitle
A novel machine learning approach for real-time determination of peptide identifiability in mass spectrometry-based proteomics experiments
Programtype2
Studentship Programs
Orgnameinp2
University of Ottawa
Cihramount2
$17,500
Cihrequipment2
$0
Programname2
Master's Award: Canada Graduate Scholarships
Keyworddelim
Analytical Chemistry; Artificial Intelligence; Artificial Neural Network; Bioinformatics; Computer Science; Machine Learning; Mass Spectrometry; Proteins; Proteomics; Supervised Learning
Abstract
Proteins have diverse functions that shape cell behaviour, making their identification of great interest to understand cellular processes. Large-scale protein identification is commonly performed through mass spectrometry-based proteomics. Prior to mass spectrometry analysis, proteins are digested into peptides to facilitate their identification. These peptides are then fragmented by the mass spectrometer, which acquires a mass spectrum of the fragments from which a software tool can identify the peptide amino acid sequence. Due to the large number of peptides in complex biological samples, mass spectrometers do not have the time to collect mass spectra for all peptides, leaving many peptides and their parent proteins unidentified. Additionally, mass spectrometers often acquire mass spectra for peptides that the analysis software will be unable to identify after the completion of the experiment. This proves a waste of resources that could otherwise be used to analyze identifiable peptides and allow the characterization of more proteins. I hypothesize that an algorithmic approach can be developed to improve mass spectrometry’s identification capabilities by predicting in real-time during an experiment whether a given peptide will be identified if the instrument collects its mass spectrum. I will design a supervised machine learning algorithm that will use peptide properties collected by the instrument prior to fragmentation to predict, in real-time during mass spectrometry analysis, which peptides are most likely to be identified if a mass spectrum is acquired for them. My model will optimize mass spectrometry resource usage to improve protein characterization in biological samples. It will yield a better understanding of the active biological processes in such samples.
Competitiondate
202312
Approvedterm2
1 yr 0 mth
Pinamesdelim
El-Azouni, Maram A
Approvedterm
1 an 0 mois
Cihrcontribution2
~ $17,500~ $0~