Method of predicting splice sites based on signal interactions

Abstract

Background: Predicting and proper ranking of canonical splice sites (SSs) is a challenging problem in bioinformatics and machine learning communities. Any progress in SSs recognition will lead to better understanding of splicing mechanism. We introduce several new approaches of combining a priori knowledge for improved SS detection. First, we design our new Bayesian SS sensor based on oligonucleotide counting. To further enhance prediction quality, we applied our new de novo motif detection tool MHMMotif to intronic ends and exons. We combine elements found with sensor information using Naive Bayesian Network, as implemented in our new tool SpliceScan.

Results: According to our tests, the Bayesian sensor outperforms the contemporary Maximum Entropy sensor for 5' SS detection. We report a number of putative Exonic (ESE) and Intronic (ISE) Splicing Enhancers found by MHMMotif tool. T-test statistics on mouse/rat intronic alignments indicates, that detected elements are on average more conserved as compared to other oligos, which supports our assumption of their functional importance. The tool has been shown to outperform the SpliceView, GeneSplicer, NNSplice, Genio and NetUTR tools for the test set of human genes. SpliceScan outperforms all contemporary ab initio gene structural prediction tools on the set of 5' UTR gene fragments.

Conclusion: Designed methods have many attractive properties, compared to existing approaches. Bayesian sensor, MHMMotif program and SpliceScan tools are freely available on our web site.

Reviewers: This article was reviewed by Manyuan Long, Arcady Mushegian and Mikhail Gelfand.

Figure 1

Consensus motifs for donor and acceptor SSs. Y-axis indicates the strength of base composition bias based on information content.

Figure 2

Models of exon definition and ESS-ESE interaction.

Figure 3

Bayesian sensor histograms produced for 5' SS and 3' SS signals on the test set of 250 human genes.

Figure 4

ROC diagrams for Donor and Acceptor signals.

Figure 5

Study of cross-correlation performance dependency.

Figure 6

Study of sensor performance based on learning set size.

Figure 7

ISE motifs found in vicinity of 3' SS (Figures 7(a)-7(g)) and 5' SS (Figures 7(h)-7(m)).

Figure 8

ESE motifs repetitively detected in our MHHMotif runs.

Figure 9

ROC diagrams for Donor and Acceptor applications.

Figure 10

Blocks placement within consensus.

Figure 11

LOD diagram for GGG signal, reported as an ISE (Figures 11(a), 11(b)). LOD diagram for 9G8 signals, reported as an ESE (Figures 11(c),11(d))

Figure 12

In our HHMM motif model B denotes background state – equiprobable emission of A, C, G, T. X is a special marker for sticky end handling to ensure proper convolution patterns. Sticky end of 10 X's is automatically added to every sample sequence by our tool.

Figure 13

MHMM model we use, where μ₁, ..., μ_K are the mixing proportions of components such that .

Figure 14

Donor and acceptor histograms approximated with a mixture of Beta distributions.

Figure 15

Example of ISE signal interactions.

Figure 16

LOD diagrams for Donor and Acceptor signal interactions.