Limits...
Machine learning for regulatory analysis and transcription factor target prediction in yeast.

Holloway DT, Kon M, Delisi C - Syst Synth Biol (2007)

Bottom Line: For the purpose of illustration we discuss several results, including biochemical pathway predictions for Gcn4 and Rap1.Future work on this method will focus on increasing the accuracy and quality of predictions using feature reduction and clustering strategies.Since predictions have been made on only 104 TFs in yeast, new classifiers will be built for the remaining 100 factors which have available binding data.

View Article: PubMed Central - PubMed

Affiliation: Molecular Biology Cell Biology and Biochemistry, Boston University, Boston, MA, 02215, USA, dth128@bu.edu.

ABSTRACT
High throughput technologies, including array-based chromatin immunoprecipitation, have rapidly increased our knowledge of transcriptional maps-the identity and location of regulatory binding sites within genomes. Still, the full identification of sites, even in lower eukaryotes, remains largely incomplete. In this paper we develop a supervised learning approach to site identification using support vector machines (SVMs) to combine 26 different data types. A comparison with the standard approach to site identification using position specific scoring matrices (PSSMs) for a set of 104 Saccharomyces cerevisiae regulators indicates that our SVM-based target classification is more sensitive (73 vs. 20%) when specificity and positive predictive value are the same. We have applied our SVM classifier for each transcriptional regulator to all promoters in the yeast genome to obtain thousands of new targets, which are currently being analyzed and refined to limit the risk of classifier over-fitting. For the purpose of illustration we discuss several results, including biochemical pathway predictions for Gcn4 and Rap1. For both transcription factors SVM predictions match well with the known biology of control mechanisms, and possible new roles for these factors are suggested, such as a function for Rap1 in regulating fermentative growth. We also examine the promoter melting temperature curves for the targets of YJR060W, and show that targets of this TF have potentially unique physical properties which distinguish them from other genes. The SVM output automatically provides the means to rank dataset features to identify important biological elements. We use this property to rank classifying k-mers, thereby reconstructing known binding sites for several TFs, and to rank expression experiments, determining the conditions under which Fhl1, the factor responsible for expression of ribosomal protein genes, is active. We can see that targets of Fhl1 are differentially expressed in the chosen conditions as compared to the expression of average and negative set genes. SVM-based classifiers provide a robust framework for analysis of regulatory networks. Processing of classifier outputs can provide high quality predictions and biological insight into functions of particular transcription factors. Future work on this method will focus on increasing the accuracy and quality of predictions using feature reduction and clustering strategies. Since predictions have been made on only 104 TFs in yeast, new classifiers will be built for the remaining 100 factors which have available binding data.

No MeSH data available.


Related in: MedlinePlus

Random vs. combined classifiers. (a) Distribution of F1 scores for normal random classifiers, (b) the same distribution on classifiers made from 26 dataset combinations for all TFs. (c) Sensitivity distribution for normal random classifiers and (d) the sensitivity distribution for the 26 dataset classifiers for all TFs
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2533145&req=5

Fig4: Random vs. combined classifiers. (a) Distribution of F1 scores for normal random classifiers, (b) the same distribution on classifiers made from 26 dataset combinations for all TFs. (c) Sensitivity distribution for normal random classifiers and (d) the sensitivity distribution for the 26 dataset classifiers for all TFs

Mentions: The dynamics of the individual classifiers can also be examined based on distributions of sensitivity and F1 score as compared to the random classifier. Figure 4a, c show the distribution of F1 score and sensitivity, respectively, for normal random data. Figure 4b, d show the same distributions but for actual data (26 method combination with tangent weights). The sensitivities and F1 scores for actual data have distributions heavily shifted to the right as opposed to those for random data. Although the majority of classifiers are comparatively good, several TFs have poor performance, something which warrants further inspection. There are four classifiers for which the F1 score and sensitivity are zero (YHL020C, YNL139C, YER068W, and YER161C). These factors have comparatively few known targets compared to others. On average these four TFs have 10 targets each (one of them has only three positives) in their training sets compared to an average of 88 targets for most regulators. This low number of positive examples is likely the cause of the poor performance. Figure 5 shows a plot of sensitivity vs. TF sorted by increasing number of positives for all classifiers. The general trend shows that classifiers having more positives give better performance.Fig. 4


Machine learning for regulatory analysis and transcription factor target prediction in yeast.

Holloway DT, Kon M, Delisi C - Syst Synth Biol (2007)

Random vs. combined classifiers. (a) Distribution of F1 scores for normal random classifiers, (b) the same distribution on classifiers made from 26 dataset combinations for all TFs. (c) Sensitivity distribution for normal random classifiers and (d) the sensitivity distribution for the 26 dataset classifiers for all TFs
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2533145&req=5

Fig4: Random vs. combined classifiers. (a) Distribution of F1 scores for normal random classifiers, (b) the same distribution on classifiers made from 26 dataset combinations for all TFs. (c) Sensitivity distribution for normal random classifiers and (d) the sensitivity distribution for the 26 dataset classifiers for all TFs
Mentions: The dynamics of the individual classifiers can also be examined based on distributions of sensitivity and F1 score as compared to the random classifier. Figure 4a, c show the distribution of F1 score and sensitivity, respectively, for normal random data. Figure 4b, d show the same distributions but for actual data (26 method combination with tangent weights). The sensitivities and F1 scores for actual data have distributions heavily shifted to the right as opposed to those for random data. Although the majority of classifiers are comparatively good, several TFs have poor performance, something which warrants further inspection. There are four classifiers for which the F1 score and sensitivity are zero (YHL020C, YNL139C, YER068W, and YER161C). These factors have comparatively few known targets compared to others. On average these four TFs have 10 targets each (one of them has only three positives) in their training sets compared to an average of 88 targets for most regulators. This low number of positive examples is likely the cause of the poor performance. Figure 5 shows a plot of sensitivity vs. TF sorted by increasing number of positives for all classifiers. The general trend shows that classifiers having more positives give better performance.Fig. 4

Bottom Line: For the purpose of illustration we discuss several results, including biochemical pathway predictions for Gcn4 and Rap1.Future work on this method will focus on increasing the accuracy and quality of predictions using feature reduction and clustering strategies.Since predictions have been made on only 104 TFs in yeast, new classifiers will be built for the remaining 100 factors which have available binding data.

View Article: PubMed Central - PubMed

Affiliation: Molecular Biology Cell Biology and Biochemistry, Boston University, Boston, MA, 02215, USA, dth128@bu.edu.

ABSTRACT
High throughput technologies, including array-based chromatin immunoprecipitation, have rapidly increased our knowledge of transcriptional maps-the identity and location of regulatory binding sites within genomes. Still, the full identification of sites, even in lower eukaryotes, remains largely incomplete. In this paper we develop a supervised learning approach to site identification using support vector machines (SVMs) to combine 26 different data types. A comparison with the standard approach to site identification using position specific scoring matrices (PSSMs) for a set of 104 Saccharomyces cerevisiae regulators indicates that our SVM-based target classification is more sensitive (73 vs. 20%) when specificity and positive predictive value are the same. We have applied our SVM classifier for each transcriptional regulator to all promoters in the yeast genome to obtain thousands of new targets, which are currently being analyzed and refined to limit the risk of classifier over-fitting. For the purpose of illustration we discuss several results, including biochemical pathway predictions for Gcn4 and Rap1. For both transcription factors SVM predictions match well with the known biology of control mechanisms, and possible new roles for these factors are suggested, such as a function for Rap1 in regulating fermentative growth. We also examine the promoter melting temperature curves for the targets of YJR060W, and show that targets of this TF have potentially unique physical properties which distinguish them from other genes. The SVM output automatically provides the means to rank dataset features to identify important biological elements. We use this property to rank classifying k-mers, thereby reconstructing known binding sites for several TFs, and to rank expression experiments, determining the conditions under which Fhl1, the factor responsible for expression of ribosomal protein genes, is active. We can see that targets of Fhl1 are differentially expressed in the chosen conditions as compared to the expression of average and negative set genes. SVM-based classifiers provide a robust framework for analysis of regulatory networks. Processing of classifier outputs can provide high quality predictions and biological insight into functions of particular transcription factors. Future work on this method will focus on increasing the accuracy and quality of predictions using feature reduction and clustering strategies. Since predictions have been made on only 104 TFs in yeast, new classifiers will be built for the remaining 100 factors which have available binding data.

No MeSH data available.


Related in: MedlinePlus