Richard J. Mural

Discovering and understanding genes in human DNA sequence using GRAIL.

Publisher Summary This chapter describes the merits of several basic analysis and annotation paradigms available to the user within the Gene Relationships Across Implicated Loci (GRAIL) and genQuest tool suite, ranging from totally automated processing to the interactive analysis of large genomic regions. It also demonstrates what biological insights can be gained or missed in the analysis of several types of DNAs. The tools and techniques that are discussed aid in this analysis by identifying biologically relevant features in the sequence, and the chapter provides some insight into their potential function. With modest effort, an investigator can greatly enrich the value of the sequence under study by including descriptions of the genes, proteins, and regulatory regions that are present. Such analysis will provide a starting point to this most exciting phase of genome research. Constructing GRAIL has also required consideration of many practical issues related to users and useful analysis procedures.

Bacterial phylogenetic diversity and a novel candidate division of two humid region, sandy surface soils

The extent of microbial community diversity in two similar sandy surface soils from Virginia and Delaware (USA) was analysed with a culture-independent small subunit ribosomal RNA (SSU rRNA) gene-based cloning approach with about 400– 700 SSU rDNA clones obtained from each sample. While there were no operational taxonomic units (OTUs) having more than three individuals, about 96 – 99% of the OTUs had only a single individual. The clones showing less than 85% similarity to the sequences in the current databases were fully sequenced. The majority of the clones (55%) had sequences that were more than 20% different from those in the current databases. About 37% of the clones differed by 15 – 20% in sequence from the database, 16% of the clones differed by 10 – 15%, and 5% of the clones differed by only 1 – 10%. Phylogenetic analysis indicated that these sequences fell into 10 of the 35 – 40 known phylogenetic divisions. Many of the clones were affiliated with Acidobacterium (35%). While a substantial portion of the clones belong to alpha (24%) and beta (12%) Proteobacteria, a few of them were affiliated with delta (6%) and gamma (3%) Proteobacteria. About 6% of the clones belong to Planctomycetes, and 4% of the clones were related to gram-positive bacteria. About 4% of clones were related to other bacterial divisions, including Cytophaga, Green sulfur bacteria, Nitrospira, OP10, and Verrucomicrobia. Eight sequences had no specific association with any of the known divisions or candidate divisions and were phylogenetically divided into three novel division level groups, named AD1, AD2 and AD3. Candidate division AD1 represented by six clones (4%) was found in both sites and consisted of two subdivisions. The community structures were similar between these two widely separated, sandy, oligotrophic, surface soils under grass vegetation in a temperate, humid climate but somewhat dissimilar to community structures revealed in similar studies in other types of soil habitats. Published by Elsevier Science Ltd.

An artificial intelligence approach to DNA sequence feature recognition

The ultimate goal of the Human Genome project is to extract the biologically relevant information recorded in the estimated 100,000 genes encoded by the 3 x 10(9) bases of the human genome. This necessitates development of reliable computer-based methods capable of analysing and correctly identifying genes in the vast amounts of DNA-sequence data generated. Such tools may save time and labour by simplifying, for example, screening of cDNA libraries. They may also facilitate the localization of human disease genes by identifying candidate genes in promising regions of anonymous DNA sequence.

GRAIL: an integrated artificial intelligence system for gene recognition and interpretation

The development of an integrated artificial intelligence system, GRAIL (gene recognition and analysis Internet link) is described. This system uses a combination of a multi-sensor/neural network, expert system, and parallel search tools to recognize and interpret genes in DNA sequences. A simple electronic mail (E-mail) interface makes the system accessible through Internet. The strength of the system in recognizing and interpreting genes in DNA sequences and the simple E-mail interface have already attracted more than 150 users. The success of the system is largely due to the multi-sensor/neural network approach and the integration of several AI tools. The modular development and flexible framework have made it easier to incorporate new knowledge and tools into the existing system.<<ETX>>

Use of neural networks for prediction of graft failure following liver transplantation

Liver transplantation is a well-established therapeutic option for patients with end-stage liver disease. However, up to 20% of transplanted livers fail to have adequate function initially, and at least half of those will eventually fail. Accurate, early prediction of outcome may ameliorate this situation by encouraging retransplantation before the patients condition becomes irreversible. In this study, clinical information was gathered prospectively for 295 patients who underwent liver transplantation at the University of Pittsburgh Medical Center, and was divided into sets. The feedforward, fully connected neural networks had 7 or 8 inputs, a single hidden layer consisting of 3 nodes and a single output node (failure=1, success=0). The networks were trained with data from a randomly selected subset of 240 patients while the remaining 55 patients made up the test set. The network was trained using a standard backpropagation algorithm. Training was assessed by testing the ability of the network to correctly predict the outcome of the 55 patients in the test set. The accuracy of prediction by the neural network improved each day and so by day 5, 98% of the graft survivors in the test set were correctly predicted while 88% of graft failures in the test set were correctly predicted.<<ETX>>

Essentiality of Glu-48 of ribulose bisphosphate carboxylase/oxygenase as demonstrated by site-directed mutagenesis.

Previous reports provide indirect evidence for the presence of Glu-48 at the active site of ribulose bisphosphate carboxylase/oxygenase from Rhodospirillum rubrum. This possibility has been examined directly by replacement of Glu-48 with glutamine via site-directed mutagenesis. This single amino acid substitution does not prevent subunit association or ligand binding. However, the Glu-48 mutant is severely deficient in catalytic activity, exhibiting a kcat only 0.05% that of wild-type enzyme. These results demonstrate that Glu-48 is positioned at the active site and suggest that it serves a functional role. In conjunction with previous studies, the discovery of essentiality of Glu-48 argues that the active site is located at an interface between subunits.

CURRENT STATUS OF COMPUTATIONAL GENE FINDING : A PERSPECTIVE

Publisher Summary This chapter discusses the present status of computational approaches to gene finding and its utility. It must be stressed that a gene predicted by computational methods must be viewed as a hypothesis subject to experimental verification. In some cases, there may be strong evidence to support the model, the transcript of the model gene is an exact match to the sequence of a cDNA in GenBank, for example, but in many cases, confirmation of the prediction will come only from the generation of further data. Another distinction that needs to be made is between finding a gene and providing a complete and accurate description of it. Often, the correct prediction of a single exon is sufficient to identify, and properly locate, a gene in a genomic sequence even if its complete intron-exon structure is not properly represented. The chapter discusses some of the limitations of the present approaches and their causes.

Protein structure prediction using hybrid AI methods

Describes a new approach for predicting protein structures based on artificial intelligence methods and genetic algorithms. We combine nearest neighbor searching algorithms, neural networks, heuristic rules and genetic algorithms to form an integrated system to predict protein structures from their primary amino acid sequences. First, we describe our methods and how they are integrated, and then apply our methods to several protein sequences. The results are very close to the real structures obtained by crystallography. Parallel genetic algorithms are also implemented.<<ETX>>

Examination of subunit interactions at the active site of ribulose 1,5-bisphosphate carboxylase/oxygenase fromRhodospirillum rubrum by hybridization of site-directed mutants

The two active sites of homodimeric ribulose bisphosphate carboxylase/oxygenase fromRhodospirillum rubrum are constituted by interacting domains of adjacent subunits, in which residues from each are required for catalytic activity. Active-site residues include Lys-166 of one domain and Glu-48 of the interacting domain from the adjacent subunit. Whereas all substitutions for Lys-166, introduced by site-directed mutagenesis, abolished catalytic activity, only a negatively charged residue (e.g., aspartic acid) resulted in the disruption of the subunit interactions (Lee et al., 1987). This disruption could result from improper folding of the individual polypeptide chains or to more localized effects (e.g., charge-charge repulsion due to proximal negative charges of Asp-166 and Glu-48 of adjacent domains or conformational changes restricted to a single domain). To address these questions, we have examined the ability of the Asp-166 mutant subunit to associate with a mutant subunit in which the negatively charged Glu-48 has been replaced by the neutral glutaminyl residue. Coexpression inEscherichia coli of the genes for both mutant subunits results in formation of a catalytically active hybrid, despite the absence of activity when either gene is expressed individually. Isolation and characterization of the hybrid show that it is composed of one Asp-166 subunit and one Gln-48 subunit, presumably with only one functional active site per dimeric molecule. This association of dissimilar subunits shows that introduction of a negative charge at position 166 does not lead to overall distortion of subunit conformation. In contrast to the wild-type enzyme, the hybrid dissociates spontaneously at low protein concentration but is stablized by elevated ionic strengths or by glycerol.

Sequence comparison on a cluster of workstations using the PVM system

Sequence comparison is one of the most important tools in molecular biology research. As the amount of DNA data increases rapidly, efficient sequence comparison algorithms are essential in studying newly discovered sequences. We have implemented a distributed sequence comparison algorithm by T.F. Smith and M. Waterman (1981) on a cluster of workstations using the PVM paradigm. This implementation has achieved similar performance to the Intel iPSC/860 hypercube, a massively parallel computer. The distributed Smith-Waterman algorithm serves as a search tool for two Internet servers GRAIL and GENQUEST. This paper describes the implementation and the performance of the algorithm.<<ETX>>