Limits...
A machine learning approach to explore the spectra intensity pattern of peptides using tandem mass spectrometry data.

Zhou C, Bowler LD, Feng J - BMC Bioinformatics (2008)

Bottom Line: The predictions include not only the mean values of spectra intensity but also the variances that can be used to tolerate noises and system biases within experimental MS/MS spectra.The features with significant influence can be used in turn to predict spectra intensities.Such information can help develop more reliable algorithms for peptide and protein identification.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Computer Science and Mathematics, University of Warwick, Coventry CV4 7AL, UK. joecong2@yahoo.co.uk

ABSTRACT

Background: A better understanding of the mechanisms involved in gas-phase fragmentation of peptides is essential for the development of more reliable algorithms for high-throughput protein identification using mass spectrometry (MS). Current methodologies depend predominantly on the use of derived m/z values of fragment ions, and, the knowledge provided by the intensity information present in MS/MS spectra has not been fully exploited. Indeed spectrum intensity information is very rarely utilized in the algorithms currently in use for high-throughput protein identification.

Results: In this work, a Bayesian neural network approach is employed to analyze ion intensity information present in 13878 different MS/MS spectra. The influence of a library of 35 features on peptide fragmentation is examined under different proton mobility conditions. Useful rules involved in peptide fragmentation are found and subsets of features which have significant influence on fragmentation pathway of peptides are characterised. An intensity model is built based on the selected features and the model can make an accurate prediction of the intensity patterns for given MS/MS spectra. The predictions include not only the mean values of spectra intensity but also the variances that can be used to tolerate noises and system biases within experimental MS/MS spectra.

Conclusion: The intensity patterns of fragmentation spectra are informative and can be used to analyze the influence of various characteristics of fragmented peptides on their fragmentation pathway. The features with significant influence can be used in turn to predict spectra intensities. Such information can help develop more reliable algorithms for peptide and protein identification.

Show MeSH

Related in: MedlinePlus

Reduction of training errors in the feature selection phase. Features are reduced according to their relevance to the fragmentation process (Figure 2). The X-axis represents the number of features being reduced and the Y-axis represents the average training error in percentage over 100 training times counted in percentage. The training error increases significantly when 23 less relevant features are removed, as indicated by the red arrow. It is then suggested that at most 22 features could be eliminated.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC2529326&req=5

Figure 4: Reduction of training errors in the feature selection phase. Features are reduced according to their relevance to the fragmentation process (Figure 2). The X-axis represents the number of features being reduced and the Y-axis represents the average training error in percentage over 100 training times counted in percentage. The training error increases significantly when 23 less relevant features are removed, as indicated by the red arrow. It is then suggested that at most 22 features could be eliminated.

Mentions: Having determined the irrelevance scores for all features examined, a new feature set can be defined containing only those found to markedly influence peptide fragmentation. To this end, we sequentially discarded the features with highest scores as listed in Figure 2, and then retrained the network with the reduced feature set. Comparison of the training results for all networks is illustrated in Figure 4. Taking non-mobile peptides as an example, the training error increases significantly when 23 less relevant features are removed, indicating that at most 22 features can be removed. The remaining features are indicated with filled circles in Figure 2.


A machine learning approach to explore the spectra intensity pattern of peptides using tandem mass spectrometry data.

Zhou C, Bowler LD, Feng J - BMC Bioinformatics (2008)

Reduction of training errors in the feature selection phase. Features are reduced according to their relevance to the fragmentation process (Figure 2). The X-axis represents the number of features being reduced and the Y-axis represents the average training error in percentage over 100 training times counted in percentage. The training error increases significantly when 23 less relevant features are removed, as indicated by the red arrow. It is then suggested that at most 22 features could be eliminated.
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC2529326&req=5

Figure 4: Reduction of training errors in the feature selection phase. Features are reduced according to their relevance to the fragmentation process (Figure 2). The X-axis represents the number of features being reduced and the Y-axis represents the average training error in percentage over 100 training times counted in percentage. The training error increases significantly when 23 less relevant features are removed, as indicated by the red arrow. It is then suggested that at most 22 features could be eliminated.
Mentions: Having determined the irrelevance scores for all features examined, a new feature set can be defined containing only those found to markedly influence peptide fragmentation. To this end, we sequentially discarded the features with highest scores as listed in Figure 2, and then retrained the network with the reduced feature set. Comparison of the training results for all networks is illustrated in Figure 4. Taking non-mobile peptides as an example, the training error increases significantly when 23 less relevant features are removed, indicating that at most 22 features can be removed. The remaining features are indicated with filled circles in Figure 2.

Bottom Line: The predictions include not only the mean values of spectra intensity but also the variances that can be used to tolerate noises and system biases within experimental MS/MS spectra.The features with significant influence can be used in turn to predict spectra intensities.Such information can help develop more reliable algorithms for peptide and protein identification.

View Article: PubMed Central - HTML - PubMed

Affiliation: Department of Computer Science and Mathematics, University of Warwick, Coventry CV4 7AL, UK. joecong2@yahoo.co.uk

ABSTRACT

Background: A better understanding of the mechanisms involved in gas-phase fragmentation of peptides is essential for the development of more reliable algorithms for high-throughput protein identification using mass spectrometry (MS). Current methodologies depend predominantly on the use of derived m/z values of fragment ions, and, the knowledge provided by the intensity information present in MS/MS spectra has not been fully exploited. Indeed spectrum intensity information is very rarely utilized in the algorithms currently in use for high-throughput protein identification.

Results: In this work, a Bayesian neural network approach is employed to analyze ion intensity information present in 13878 different MS/MS spectra. The influence of a library of 35 features on peptide fragmentation is examined under different proton mobility conditions. Useful rules involved in peptide fragmentation are found and subsets of features which have significant influence on fragmentation pathway of peptides are characterised. An intensity model is built based on the selected features and the model can make an accurate prediction of the intensity patterns for given MS/MS spectra. The predictions include not only the mean values of spectra intensity but also the variances that can be used to tolerate noises and system biases within experimental MS/MS spectra.

Conclusion: The intensity patterns of fragmentation spectra are informative and can be used to analyze the influence of various characteristics of fragmented peptides on their fragmentation pathway. The features with significant influence can be used in turn to predict spectra intensities. Such information can help develop more reliable algorithms for peptide and protein identification.

Show MeSH
Related in: MedlinePlus