Can Clustal-style progressive pairwise alignment of multiple sequences be used in RNA secondary structure prediction?

Abstract

Background: In ribonucleic acid (RNA) molecules whose function depends on their final, folded three-dimensional shape (such as those in ribosomes or spliceosome complexes), the secondary structure, defined by the set of internal basepair interactions, is more consistently conserved than the primary structure, defined by the sequence of nucleotides.

Results: The research presented here investigates the possibility of applying a progressive, pairwise approach to the alignment of multiple RNA sequences by simultaneously predicting an energy-optimized consensus secondary structure. We take an existing algorithm for finding the secondary structure common to two RNA sequences, Dynalign, and alter it to align profiles of multiple sequences. We then explore the relative successes of different approaches to designing the tree that will guide progressive alignments of sequence profiles to create a multiple alignment and prediction of conserved structure.

Conclusion: We have found that applying a progressive, pairwise approach to the alignment of multiple ribonucleic acid sequences produces highly reliable predictions of conserved basepairs, and we have shown how these predictions can be used as constraints to improve the results of a single-sequence structure prediction algorithm. However, we have also discovered that the amount of detail included in a consensus structure prediction is highly dependent on the order in which sequences are added to the alignment (the guide tree), and that if a consensus structure does not have sufficient detail, it is less likely to provide useful constraints for the single-sequence method.

Figure 1

Boxplots showing the range of sensitivity and positive predictive value statistics for mfold optimum predictions on the two test data sets.

Figure 2

Boxplots showing the range of sensitivity and positive predictive value statistics for all pairwise alignments of the tRNA sequences, at each gap penalty value tested.

Figure 3

Boxplots showing the range of sensitivity and positive predictive value statistics for all pairwise alignments of the 5S rRNA sequences, at each gap penalty value tested.

Figure 4

Sensitivity versus positive predictive value (PPV) for all predicted tRNA structures from the pairwise alignments and all stages of the randomized progressive alignments, with the gap penalty used and the number of sequences in the alignment indicated by the colour and size of the points, respectively.

Figure 5

Sensitivity versus positive predictive value (PPV) for all predicted 5S rRNA structures from the pairwise alignments and all stages of the randomized progressive alignments, with the gap penalty used and the number of sequences in the alignment indicated by the colour and size of the points, respectively.

Figure 6

Boxplots showing the range in sensitivity and positive predictive value (PPV) for the twelve predicted tRNA structures from the final alignment and consensus predictions of all runs with gap penalty ≥ 2.0 kcal/mol, defined by the algorithm used to build the guide tree. NJ is neighbor-joining, CW is Clustal W, B1, B2 and B3 represent the 3 runs with balanced guide trees, while L1, L2 and L3 represent the 3 runs with linear guide trees. Undefined PPV values are plotted as zero.

Figure 7

Boxplots showing the range in sensitivity and positive predictive value (PPV) for the twelve predicted 5S rRNA structures from the final alignment and consensus predictions of all runs with gap penalty ≥ 2.0 kcal/mol, defined by the algorithm used to build the guide tree. NJ is neighbor-joining, CW is Clustal W, B1, B2 and B3 represent the 3 runs with balanced guide trees, while L1, L2 and L3 represent the 3 runs with linear guide trees. Undefined PPV values are plotted as zero.

Figure 8

The phylogeny predicted by Clustal W for the tRNA sequences (a), and the guide tree created by neighbour-joining the scores from the pairwise alignments with gap penalty of 4 kcal/mol (b). Images produced by drawgram, from the Phylip package of programs [30].

Figure 9

The phylogeny predicted by Clustal W for the 5S rRNA sequences (a), and the guide tree created by neighbour-joining the scores from the pairwise alignments with gap penalty of 4 kcal/mol (b). Images produced by drawgram, from the Phylip package of programs [30].

Figure 10

The sensitivity, positive predictive value (PPV), and Matthew's Correlation Coefficient (MCC) statistics for the unconstrained mfold optimum structure, selected consensus structures, and the mfold optimum prediction when the consensus structure is used as a forced constraint ('refolded'), for each tRNA sequence. (a) Gap penalty of 4.0 kcal/mol used to generate consensus alignment. (b) Gap penalty of 6.0 kcal/mol used to generate consensus alignment.

Figure 11

The sensitivity, positive predictive value (PPV), and Matthew's Correlation Coefficient (MCC) statistics for the unconstrained mfold optimum structure, selected consensus structure, and the mfold optimum prediction when the consensus structure is used as a forced constraint ('refolded'), for each 5S rRNA sequence.

Figure 12

The reference secondary structure predicted for 5S rRNA sequence V00336 (a), the structure predicted by mfold as the unconstrained optimum (b), the conserved structure predicted by the nearest-neighbour consensus algorithm with gap penalty 4 or 6 kcal/mol (c), and the structure predicted by mfold as the optimum, when it was forced to include the consensus basepairs (d). Images produced by the sir graph utility of the mfold program [24].