A non-independent energy-based multiple sequence alignment improves prediction of transcription factor binding sites.

MOTIVATION: Multiple sequence alignments (MSAs) are usually scored under the assumption that the sequences being aligned have evolved by common descent. Consequently, the differences between sequences reflect the impact of insertions, deletions and mutations. However, non-coding DNA binding sequences, such as transcription factor binding sites (TFBSs), are frequently not related by common descent, and so the existing alignment scoring methods are not well suited for aligning such sequences. RESULTS: We present a novel multiple MSA methodology that scores TFBS DNA sequences by including the interdependence of neighboring bases. We introduced two variants supported by different underlying null hypotheses, one statistically and the other thermodynamically generated. We assessed the alignments through their performance in TFBS prediction; both methods show considerable improvements when compared with standard MSA algorithms. Moreover, the thermodynamically generated null hypothesis outperforms the statistical one due to improved stability in the base stacking free energy of the alignment. The thermodynamically generated null hypothesis method can be downloaded from http://sourceforge.net/projects/msa-edna/. CONTACT: dov.stekel@nottingham.ac.uk. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The genotype of RAET1L (ULBP6), a ligand for human NKG2D (KLRK1), markedly influences the clinical outcome of allogeneic stem cell transplantation.

NKG2D (KLRK1) is an activating receptor on natural killer (NK) and T-cells and binds a diverse panel of polymorphic ligands encoded by the MIC and RAET1 gene families. We studied the clinical importance of retinoic acid early transcript-1 (RAET1) polymorphism in allogeneic stem cell transplantation (SCT) by determining the frequency of 18 single nucleotide polymorphisms (SNPs) and individual RAET1 alleles in 371 patient-donor pairs and relating this to clinical outcome. A strong association was observed between the presence of five SNPs within the patient RAET1L (ULBP6) gene and relapse-free survival and overall survival. Two common alleles of RAET1L were determined and the presence of the protective RAET1L*02 allele in the patient was associated with a relapse-free survival of 44% at 8 years compared with just 25% in patients who lacked a RAET1L*02 allele (P < 0·001). Overall survival at this time was 55% in those with RAET1L*02 allele compared to 39% in patients who lacked a RAET1L*02 allele (P = 0·003). These novel findings indicate a critical role for NKG2D-RAET1L interactions in determining SCT clinical outcome and show RAET1L may have an important influence on regulating the strength of the alloreactive immune response. The data will be of value in guiding the development of future transplant therapy protocols.

Inclusion of neighboring base interdependencies substantially improves genome-wide prokaryotic transcription factor binding site prediction.

Prediction of transcription factor binding sites is an important challenge in genome analysis. The advent of next generation genome sequencing technologies makes the development of effective computational approaches particularly imperative. We have developed a novel training-based methodology intended for prokaryotic transcription factor binding site prediction. Our methodology extends existing models by taking into account base interdependencies between neighbouring positions using conditional probabilities and includes genomic background weighting. This has been tested against other existing and novel methodologies including position-specific weight matrices, first-order Hidden Markov Models and joint probability models. We have also tested the use of gapped and ungapped alignments and the inclusion or exclusion of background weighting. We show that our best method enhances binding site prediction for all of the 22 Escherichia coli transcription factors with at least 20 known binding sites, with many showing substantial improvements. We highlight the advantage of using block alignments of binding sites over gapped alignments to capture neighbouring position interdependencies. We also show that combining these methods with ChIP-on-chip data has the potential to further improve binding site prediction. Finally we have developed the ungapped likelihood under positional background platform: a user friendly website that gives access to the prediction method devised in this work.