Search

Abstract Background microRNAs (miRNAs) are tiny endogenous RNAs that have been discovered in animals and plants, and direct the post-transcriptional regulation of target mRNAs for degradation or translational repression via binding to the 3'UTRs and the coding exons. To gain insight into the biological role of miRNAs, it is essential to identify the full repertoire of mRNA targets (target genes). A number of computer programs have been developed for miRNA-target prediction. These programs essentially focus on potential binding sites in 3'UTRs, which are recognized by miRNAs according to specific base-pairing rules. Results Here, we introduce a novel method for miRNA-target prediction that is entirely independent of existing approaches. The method is based on the hypothesis that transcription of a miRNA and its target genes tend to be co-regulated by common transcription factors. This hypothesis predicts the frequent occurrence of common cis-elements between promoters of a miRNA and its target genes. That is, our proposed method first identifies putative cis-elements in a promoter of a given miRNA, and then identifies genes that contain common putative cis-elements in their promoters. In this paper, we show that a significant number of common cis-elements occur in ~28% of experimentally supported human miRNA-target data. Moreover, we show that the prediction of human miRNA-targets based on our method is statistically significant. Further, we discuss the random incidence of common cis-elements, their consensus sequences, and the advantages and disadvantages of our method. Conclusions This is the first report indicating prevalence of transcriptional regulation of a miRNA and its target genes by common transcription factors and the predictive ability of miRNA-targets based on this property.

Author(s): International Human Genome Sequencing Consortium; Whitehead Institute for Biomedical Research, Center for Genome Research:; Eric S. Lander [1]; Lauren M. Linton [1]; Bruce Birren [1]; Chad Nusbaum [1]; Michael [...]

Author(s): The International Human Genome Mapping Consortium ; Washington University School of Medicine, Genome Sequencing Center: ; John D. McPherson (corresponding author) [1]; Marco Marra [1]; LaDeana Hillier [1]; Robert [...]

Although a number of bacterial gene finding programs have been developed, there is still room for improvement especially in the area of correctly detecting translation start sites. We developed a novel bacterial gene finding program named GeneHacker Plus. Like many others, it is based on a hidden Markov model (HMM) with duration. However, it is a ‘local’ model in the sense that the model starts from the translation control region and ends at the stop codon of a coding region. Multiple coding regions are identi.ed as partial paths, like local alignments in the Smith-Waterman algorithm, regardless of how they overlap. Moreover, our semiautomatic procedure for constructing the model of the translation control region allows the inclusion of an additional conserved element as well as the ribosome-binding site. We confirmed that GeneHacker Plus is one of the most accurate programs in terms of both finding potential coding regions and precisely locating translation start sites. GeneHacker Plus is also equipped with an option where the results from database homology searches are directly embedded in the HMM. Although this option does not raise the overall predictability, labeled similarity information can be of practical use. GeneHacker Plus can be accessed freely at . [ABSTRACT FROM PUBLISHER]

We developed a computer program, GeneHackerTL, which predicts the most probable translation initiation site for a given nucleotide sequence. The program requires that information be extracted from the nucleotide sequence data surrounding the translation initiation sites according to the framework of the Hidden Markov Model. Since the translation initiation sites of 72 highly abundant proteins have already been assigned on the genome of Synechocystis sp. strain PCC6803 by amino-terminal analysis, we extracted necessary information for GeneHackerTL from the nucleotide sequence data. The prediction rate of the GeneHackerTL for these proteins was estimated to be 86.1%. We then used GeneHackerTL for prediction of the translation initiation sites of 24 other proteins, of which the initiation sites were not assigned experimentally, because of the lack of a potential initiation codon at the amino-terminal position. For 20 out of the 24 proteins, the initiation sites were predicted in the upstream of their amino-terminal positions. According to this assignment, the processed regions represent a typical feature of signal peptides. We could also predict multiple translation initiation sites for a particular gene for which at least two initiation sites were experimentally detected. This program would be e.ective for the prediction of translation initiationsites of other proteins, not only in this species but also in other prokaryotes as well. [ABSTRACT FROM PUBLISHER]

Sequence patterns surrounding the translation initiation sites of Cyanobacterium were precisely analyzed by the hidden Markov model (HMM) based on the actual translation initiation sites. In a previous study, 72 actual protein coding regions and their translation initiation sites on the genome of Synechocystis sp. strain PCC6803 were determined by Sazuka et al. using protein two-dimensional electrophoresis and microsequening. In this work, we extracted the sequence patterns surrounding translation initiation sites as HMM using the computer program YEBIS. The constructed HMM could recognize all but one translation initiation site. The HMM contains an AG-rich region (5.7 bp on average), as the Shine-Dalgarno sequence exclusively contains purines, upstream of the translation initiation site (−9.7 position on average) and a CT-rich region (4.2 bp on average) just upstreamfrom the translation initiation site. In addition, we found that the second amino acid (+4,5,6) could be classified into two types, one of which had C as their second codon while another of which has a nucleotide distribution relatively similar to the distribution among amino acids in the 72 proteins. This fact corresponds well to our earlier finding that when the second nucleotide of the second amino acid of a translated protein was C, an initial methionine was processed and that otherwise the methionine was intact with high frequency. [ABSTRACT FROM PUBLISHER]

The gene-finding programs developed so far have not paid much attention to the detection of short protein coding regions (CDSs). However, the detection of short CDSs is important for the study of photosynthesis. We utilized GeneHacker, a gene-finding program based on the hidden Markov model (HMM), to detect short CDSs (from 90 to 300 bases) in a 1.0 mega contiguous sequence of cyanobacterium Synechocystis sp. strain PCC6803 which carries a complete set of genes for oxygenic photosynthesis. GeneHacker differs from other gene-finding programs based on the HMM in that it utilizes di-codon statistics as well. GeneHacker successfully detected seven out of the eight short CDSs annotated in this sequence and was clearly superior to GeneMark in this range of length. GeneHacker detected 94 potentially new CDSs, 9 of which have counterparts in the genetic databases. Four of the nine CDSs were less than 150 bases and were photosynthesis-related genes. The results show the effectiveness of GeneHacker in detecting very short CDSs corresponding to genes. [ABSTRACT FROM PUBLISHER]

We developed linear regression models which predict strength of transcriptional activity of promoters from their sequences. Intrinsic transcriptional strength data of 451 human promoter sequences in three cell lines (HEK293, MCF7 and 3T3), which were measured by systematic luciferase reporter gene assays, were used to build the models. The models sum up contributions of CG dinucleotide content and transcription factor binding sites (TFBSs) to transcriptional strength. We evaluated prediction accuracies of the models by cross validation tests and found that they have adequate ability for predicting transcriptional strength of promoters in spite of their simple formalization. We also evaluated statistical significance of the contributions and proposed a picture of regulatory code hidden in promoter sequences. That is, CG dinucleotide content and TFBSs mainly determine strength of transcriptional activity under ubiquitous and specific environments, respectively.

As a base for human transcriptome and functional genomics, we created the "full-length long Japan" (FLJ) collection of sequenced human cDNAs. We determined the entire sequence of 21,243 selected clones and found that 14,490 cDNAs (10,897 clusters) were unique to the FLJ collection. About half of them (5,416) seemed to be protein-coding. Of those, 1,999 clusters had not been predicted by computational methods. The distribution of GC content of nonpredicted cDNAs had a peak at approximately 58% compared with a peak at approximately 42%for predicted cDNAs. Thus, there seems to be a slight bias against GC-rich transcripts in current gene prediction procedures. The rest of the cDNAs unique to the FLJ collection (5,481) contained no obvious open reading frames (ORFs) and thus are candidate noncoding RNAs. About one-fourth of them (1,378) showed a clear pattern of splicing. The distribution of GC content of noncoding cDNAs was narrow and had a peak at approximately 42%, relatively low compared with that of protein-coding cDNAs.

We introduce here a novel index which precisely derives protein coding regions from cross-species genome alignments. The index is deeply related to frame recovery observed in coding sequence alignments, that is, if insertions or deletions of nucleotides causes frame shifts in coding regions, other in-dels which recover the reading frames will be often observed in the vicinity. In contrast, such frame recoveries are not observed in other conserved regions. We prepared two gene models: a model which finds gene by using sequence similarity and intrinsic gene measures (basic model), and the other model which finds gene by using frame recovery index in addition to sequence similarity and intrinsic gene measures (frame recovery model). We evaluated the prediction accuracies of the two models, and our benchmark test revealed that frame recovery model significantly improved the prediction accuracy in comparison with basic model.