A graph-theoretical approach for motif discovery in protein sequences

Identifying TF Binding Motifs from a Partial Set of Target Genes and its Application to Regulatory Network Inference

Motif identification has been one of the most widely studied problems in bioinformatics. Many methods have been developed to discover binding motifs from a large set of genes. But when the given genes are only a partial set of target genes, the statistical significance usually contains a bias towards the input. If we can identify the TF binding motif from a partial set of target genes, we can save the labor costs and resources for doing many experiments. In this paper, we propose a method MISA (Motif Identification through Segments Assembly) to identify binding motifs from a subset of target genes. By ranking and assembling the segments, MISA discovers a set of binding motifs with the best length to fit our proposed objective function. We also predict the additional target genes as an application of regulatory network inference. We compare our approach with two widely used methods MEME and AlignACE by analyzing both the quality of the binding motif and network inference. Using two model organisms S. cerevisiae and E. coli, we show that with 20 percent of the target genes (minimum sample size of 20), we can achieve a motif similarity of 82 percent with the known motifs. Our results also show that 73 percent of target genes on average can be correctly predicted without introducing many false target genes.

HH-MOTiF: de novo detection of short linear motifs in proteins by Hidden Markov Model comparisons

Short linear motifs (SLiMs) in proteins are self-sufficient functional sequences that specify interaction sites for other molecules and thus mediate a multitude of functions. Computational, as well as experimental biological research would significantly benefit, if SLiMs in proteins could be correctly predicted de novo with high sensitivity. However, de novo SLiM prediction is a difficult computational task. When considering recall and precision, the performances of published methods indicate remaining challenges in SLiM discovery. We have developed HH-MOTiF, a web-based method for SLiM discovery in sets of mainly unrelated proteins. HH-MOTiF makes use of evolutionary information by creating Hidden Markov Models (HMMs) for each input sequence and its closely related orthologs. HMMs are compared against each other to retrieve short stretches of homology that represent potential SLiMs. These are transformed to hierarchical structures, which we refer to as motif trees, for further processing and evaluation. Our approach allows us to identify degenerate SLiMs, while still maintaining a reasonably high precision. When considering a balanced measure for recall and precision, HH-MOTiF performs better on test data compared to other SLiM discovery methods. HH-MOTiF is freely available as a web-server at http://hh-motif.biochem.mpg.de.

A multi-objective imperialist competitive algorithm (MOICA) for finding motifs in DNA sequences

Motif discovery problem (MDP) is one of the well-known problems in biology which tries to find the transcription factor binding site (TFBS) in DNA sequences. In one aspect, there is not enough biological knowledge on motif sites and on the other side, the problem is NP-hard. Thus, there is not an efficient procedure capable of finding motifs in every dataset. Some algorithms use exhaustive search, which is very time-consuming for large-scale datasets. On the other side, metaheuristic procedures seem to be a good selection for finding a motif quickly that at least has some acceptable biological properties. Most of the previous methods model the problem as a single objective optimization problem; however, considering multi-objectives for modeling the problem leads to improvements in the quality of obtained motifs. Some multi-objective optimization models for MDP have tried to maximize three objectives simultaneously: Motif length, support, and similarity. In this study, the multi-objective Imperialist Competition Algorithm (ICA) is adopted for this problem as an approximation algorithm. ICA is able to simulate more exploration along the solution space, so avoids trapping into local optima. So, it promises to obtain good solutions in a reasonable time. Experimental results show that our method produces good solutions compared to well-known algorithms in the literature, according to computational and biological indicators.