The applicability of recurrent neural networks for biological sequence analysis

Selection of machine learning techniques requires a certain sensitivity to the requirements of the problem. In particular, the problem can be made more tractable by deliberately using algorithms that are biased toward solutions of the requisite kind. In this paper, we argue that recurrent neural networks have a natural bias toward a problem domain of which biological sequence analysis tasks are a subset. We use experiments with synthetic data to illustrate this bias. We then demonstrate that this bias can be exploitable using a data set of protein sequences containing several classes of subcellular localization targeting peptides. The results show that, compared with feed forward, recurrent neural networks will generally perform better on sequence analysis tasks. Furthermore, as the patterns within the sequence become more ambiguous, the choice of specific recurrent architecture becomes more critical.

Orphan nuclear receptors: therapeutic opportunities in skeletal muscle

Orphan nuclear receptors: therapeutic opportunities in skeletal muscle. Am J Physiol Cell Physiol 291: C203-C217, 2006; doi: 10.1152/ajpcell. 00476.2005.-Nuclear hormone receptors (NRs) are ligand-dependent transcription factors that bind DNA and translate physiological signals into gene regulation. The therapeutic utility of NRs is underscored by the diversity of drugs created to manage dysfunctional hormone signaling in the context of reproductive biology, inflammation, dermatology, cancer, and metabolic disease. For example, drugs that target nuclear receptors generate over $10 billion in annual sales. Almost two decades ago, gene products were identified that belonged to the NR superfamily on the basis of DNA and protein sequence identity. However, the endogenous and synthetic small molecules that modulate their action were not known, and they were denoted orphan NRs. Many of the remaining orphan NRs are highly enriched in energy-demanding major mass tissues, including skeletal muscle, brown and white adipose, brain, liver, and kidney. This review focuses on recently adopted and orphan NR function in skeletal muscle, a tissue that accounts for similar to 35% of the total body mass and energy expenditure, and is a major site of fatty acid and glucose utilization. Moreover, this lean tissue is involved in cholesterol efflux and secretes that control energy expenditure and adiposity. Consequently, muscle has a significant role in insulin sensitivity, the blood lipid profile, and energy balance. Accordingly, skeletal muscle plays a considerable role in the progression of dyslipidemia, diabetes, and obesity. These are risk factors for cardiovascular disease, which is the the foremost cause of global mortality (> 16.7 million deaths in 2003). Therefore, it is not surprising that orphan NRs and skeletal muscle are emerging as therapeutic candidates in the battle against dyslipidemia, diabetes, obesity, and cardiovascular disease.

Binding of an RNA trafficking response element to heterogeneous nuclear ribonucleoproteins A1 and A2

Heterogeneous nuclear ribonucleoprotein (hnRNP) A2 binds a 21-nucleotide myelin basic protein mRNA response element, the A2RE, and A2RE-like sequences in other localized mRNAs, and is a trans-acting factor in oligodendrocyte cytoplasmic RNA trafficking. Recombinant human hnRNPs A1 and A2 were used in a biosensor to explore interactions with A2RE and the cognate oligodeoxyribonucleotide. Both proteins have a single site that bound oligonucleotides with markedly different sequences but did not bind in the presence of heparin. Both also possess a second, specific site that bound only A2RE and was unaffected by heparin, hnRNP A2 bound A2RE in the latter site with a K-d near 50 nM, whereas the K-d for hnRNP A1 was above 10 muM. UV cross-linking assays led to a similar conclusion. Mutant A2RE sequences, that in earlier qualitative studies appeared not to bind hnRNP A2 or support RNA trafficking in oligodendrocytes, had dissociation constants above 5 muM for this protein. The two concatenated RNA recognition motifs (RRMs), but not the individual RRMs, mimicked the binding behavior of hnRNP A2. These data highlight the specificity of the interaction of A2RE with these hnRNPs and suggest that the sequence-specific A2RE-binding site on hnRNP A2 is formed by both RRMs acting in cis.

Evaluation and comparison of mammalian subcellular localization prediction methods

Background: Determination of the subcellular location of a protein is essential to understanding its biochemical function. This information can provide insight into the function of hypothetical or novel proteins. These data are difficult to obtain experimentally but have become especially important since many whole genome sequencing projects have been finished and many resulting protein sequences are still lacking detailed functional information. In order to address this paucity of data, many computational prediction methods have been developed. However, these methods have varying levels of accuracy and perform differently based on the sequences that are presented to the underlying algorithm. It is therefore useful to compare these methods and monitor their performance. Results: In order to perform a comprehensive survey of prediction methods, we selected only methods that accepted large batches of protein sequences, were publicly available, and were able to predict localization to at least nine of the major subcellular locations (nucleus, cytosol, mitochondrion, extracellular region, plasma membrane, Golgi apparatus, endoplasmic reticulum (ER), peroxisome, and lysosome). The selected methods were CELLO, MultiLoc, Proteome Analyst, pTarget and WoLF PSORT. These methods were evaluated using 3763 mouse proteins from SwissProt that represent the source of the training sets used in development of the individual methods. In addition, an independent evaluation set of 2145 mouse proteins from LOCATE with a bias towards the subcellular localization underrepresented in SwissProt was used. The sensitivity and specificity were calculated for each method and compared to a theoretical value based on what might be observed by random chance. Conclusion: No individual method had a sufficient level of sensitivity across both evaluation sets that would enable reliable application to hypothetical proteins. All methods showed lower performance on the LOCATE dataset and variable performance on individual subcellular localizations was observed. Proteins localized to the secretory pathway were the most difficult to predict, while nuclear and extracellular proteins were predicted with the highest sensitivity.

Wolbachia pipientis: intracellular infection and pathogenesis in Drosophila

Wolbachia pipientis is a vertically transmitted, obligate intracellular symbiont of arthropods. The bacterium is best known for its ability to manipulate host reproductive biology where it can induce cytoplasmic incompatibility, parthenogenesis, feminization and male-killing. In addition to the various reproductive phenotypes it generates through interaction with host reproductive tissue it is also known to infect somatic tissues. However, relatively little is known about the consequences of infection of these tissues with the exception that in some hosts Wolbachia acts as a classical mutualist and in others a pathogen, dramatically shortening adult insect lifespan. Manipulation experiments have demonstrated that the severity of Wolbachia-induced effects on the host is determined by a combination of host genotype, Wolbachia strain, host tissue localization, and interaction with the environment. The recent completion of the whole genome sequence of Wolbachia pipientis wMel strain indicates that it is likely to use a type IV secretion system to establish and maintain infection in its host. Moreover, an unusual abundance of genes encoding proteins with eukaryotic-like ankyrin repeat domains suggest a function in the various described phenotypic effects in hosts.

Algorithms for sequence analysis via mutagenesis

Despite many successes of conventional DNA sequencing methods, some DNAs remain difficult or impossible to sequence. Unsequenceable regions occur in the genomes of many biologically important organisms, including the human genome. Such regions range in length from tens to millions of bases, and may contain valuable information such as the sequences of important genes. The authors have recently developed a technique that renders a wide range of problematic DNAs amenable to sequencing. The technique is known as sequence analysis via mutagenesis (SAM). This paper presents a number of algorithms for analysing and interpreting data generated by this technique.

Unlocking hidden genomic sequence

Despite the success of conventional Sanger sequencing, significant regions of many genomes still present major obstacles to sequencing. Here we propose a novel approach with the potential to alleviate a wide range of sequencing difficulties. The technique involves extracting target DNA sequence from variants generated by introduction of random mutations. The introduction of mutations does not destroy original sequence information, but distributes it amongst multiple variants. Some of these variants lack problematic features of the target and are more amenable to conventional sequencing. The technique has been successfully demonstrated with mutation levels up to an average 18% base substitution and has been used to read previously intractable poly(A), AT-rich and GC-rich motifs.

Bellerophon: a program to detect chimeric sequences in multiple sequence alignments

Bellerophon is a program for detecting chimeric sequences in multiple sequence datasets by an adaption of partial treeing analysis. Bellerophon was specifically developed to detect 16S rRNA gene chimeras in PCR-clone libraries of environmental samples but can be applied to other nucleotide sequence alignments.

Sequence and tissue distribution of chicken cellular nucleic acid binding protein cDNA

We sequenced cDNAs coding for chicken cellular nucleic acid binding protein (CNBP). Two slightly different variations of the open reading frame were found, each of which translates into a protein with seven zinc finger domains. The longest transcript contains an in-frame insert of 3 bp. The sequence conservation between chick CNBP cDNAs with human, rat and mouse CNBP cDNAs is extreme, especially in the coding region, where the deduced amino acid sequence identity with human, rat and mouse CNBP is 99%. CNBP-like transcripts were also found in various tissues from insect, shrimp, fish and lizard. Regions with remarkable nucleotide conservation were also found in the 3' untranslated region, indicating important functions for these regions. Quantitative reverse transcription polymerase chain reaction (RT-PCR) indicated that in the chick, CNBP is present in all tissues examined in approximately equal ratios to total RNA. RT-PCR of total RNA isolated from different phyla indicate CNBP-like proteins art widespread throughout the animal kingdom. The extraordinary level of conservation suggests an important physiological role for CNBP. (C) 1997 Elsevier Science Inc.

Nucleotide sequence of the nifH gene coding for nitrogen reductase in the acetic acid bacterium Acetobacter diazotrophicus

The nifH gene sequence of the nitrogen-fixing bacterium Acetobacter diazotrophicus was determined with the use of the polymerase chain reaction and universal degenerate oligonucleotide primers. The gene shows highest pair-wise similarity to the nifH gene of Azospirillum brasilense. The phylogenetic relationships of the nifH gene sequences were compared with those inferred from 16S rRNA gene sequences. Knowledge of the sequence of the nifH gene contributes to the growing database of nifH gene sequences, and will allow the detection of Acet. diazotrophicus from environmental samples with nifH gene-based primers.

Molecular cloning and sequence of the ovine gastrin gene

A clone encoding ovine preprogastrin was isolated from a sheep genomic library. The deduced 104 amino acid sequence of ovine preprogastrin was 92% and 68% identical to the sequences of bovine and human preprogastrin, respectively. While the similarity was greatest in the gastrin-17 sequence, an unexpected similarity was also observed in the N-terminus of mature progastrin.

Bacterial expression of two human aryl sulfotransferases

The effect of replacing a single codon in the N-terminal of human aryl sulfotransferase (HAST) 1 and 3 with one that is more commonly found in E. coli genes was assessed. The pKK233-2 E. coli expression vector was employed and the polymerase chain reaction (PCR) was used to introduce the 5' nucleotide substitution, at the same time maintaining the fidelity of the amino acid sequence. The data indicates that this change had a minimal effect on protein production, subcellular localization or, in the case of HAST3, catalytic activity. In general, the pKK233-2 E. coli vector has been less than optimal for expressing human sulfotransferase cDNAs. (C) 1998 Elsevier Science Ireland Ltd. All rights reserved.

Structure and distribution of N-glycans on the S-7-allele stylar self-incompatibility ribonuclease of Nicotiana alata

S-RNases are the stylar products of the self-incompatibility (S)-locus in solanaceous plants (including Nicotiana alata), and as such, are involved in the prevention of self-pollination. All cDNA sequences of S-RNase products of functional S-alleles contain potential N-glycosylation sites, with one site being conserved in all cases, suggesting that N-glycosylation is important in self-incompatibility. In this study, we report on the structure and localization of the N-glycans on the S-7-allele RNase of N, alata, A total of nine N-glycans, belonging to the high-mannose- and xylosylated hybrid-classes, were identified and characterized by a combination of electrospray-ionization mass-spectrometry (ESI-MS), H-1-NMR spectroscopy, and methylation analyses. The glycosylation pattern of individual glycosylation sites was determined by ESI-MS of the glycans released from isolated chymotryptic glycopeptides, All three N-glycosylation sites showed microheterogeneity and each had a unique complement of N-glycans, The N-glycosylation pattern of the S-7-RNase is significantly different to those of the S-1- and S-2-RNases.