Exploiting the past and the future in protein secondary structure prediction.
(9/288)
MOTIVATION: Predicting the secondary structure of a protein (alpha-helix, beta-sheet, coil) is an important step towards elucidating its three-dimensional structure, as well as its function. Presently, the best predictors are based on machine learning approaches, in particular neural network architectures with a fixed, and relatively short, input window of amino acids, centered at the prediction site. Although a fixed small window avoids overfitting problems, it does not permit capturing variable long-rang information. RESULTS: We introduce a family of novel architectures which can learn to make predictions based on variable ranges of dependencies. These architectures extend recurrent neural networks, introducing non-causal bidirectional dynamics to capture both upstream and downstream information. The prediction algorithm is completed by the use of mixtures of estimators that leverage evolutionary information, expressed in terms of multiple alignments, both at the input and output levels. While our system currently achieves an overall performance close to 76% correct prediction--at least comparable to the best existing systems--the main emphasis here is on the development of new algorithmic ideas. AVAILABILITY: The executable program for predicting protein secondary structure is available from the authors free of charge. CONTACT: [email protected], [email protected], [email protected], [email protected]. (+info)
Polymer chromosome models and Monte Carlo simulations of radiation breaking DNA.
(10/288)
MOTIVATION: Chromatin breakage by ionizing radiation is relevant to studies of carcinogenesis, tumor radiotherapy, biodosimetry and molecular biology. This article focuses on computer analysis of chromosome irradiation in mammlian cells. METHODS: Polymer physics and Monte Carlo numerical methods are used to develop a coarse-grained computational approach. Chromatin is modeled as a random walk on a cubic lattice, and the radiation tracks hitting the chromatin are modeled as straight lines hitting lattice sites. Each track can make a cluster of DSBs on a chromosome. RESULTS: The results obtained replace conjectured DNA fragment-size distribution functions in the recently developed RLC formalism by more mechanistically motivated distributions. The discrete lattice algorithm reproduces features of current radiation experiments relevant to chromatin on large scales. It approximates the continuous formalism and experimental data with adequate precision. It was also found that assuming either fixed chromatin with correlations among different clusters of DSBs or moving chromatin with no such correlations gives virtually identical numerical predictions. (+info)
MaxSub: an automated measure for the assessment of protein structure prediction quality.
(11/288)
MOTIVATION: Evaluating the accuracy of predicted models is critical for assessing structure prediction methods. Because this problem is not trivial, a large number of different assessment measures have been proposed by various authors, and it has already become an active subfield of research (Moult et al. (1997,1999) and CAFASP (Fischer et al. 1999) prediction experiments have demonstrated that it has been difficult to choose one single, 'best' method to be used in the evaluation. Consequently, the CASP3 evaluation was carried out using an extensive set of especially developed numerical measures, coupled with human-expert intervention. As part of our efforts towards a higher level of automation in the structure prediction field, here we investigate the suitability of a fully automated, simple, objective, quantitative and reproducible method that can be used in the automatic assessment of models in the upcoming CAFASP2 experiment. Such a method should (a) produce one single number that measures the quality of a predicted model and (b) perform similarly to human-expert evaluations. RESULTS: MaxSub is a new and independently developed method that further builds and extends some of the evaluation methods introduced at CASP3. MaxSub aims at identifying the largest subset of C(alpha) atoms of a model that superimpose 'well' over the experimental structure, and produces a single normalized score that represents the quality of the model. Because there exists no evaluation method for assessment measures of predicted models, it is not easy to evaluate how good our new measure is. Even though an exact comparison of MaxSub and the CASP3 assessment is not straightforward, here we use a test-bed extracted from the CASP3 fold-recognition models. A rough qualitative comparison of the performance of MaxSub vis-a-vis the human-expert assessment carried out at CASP3 shows that there is a good agreement for the more accurate models and for the better predicting groups. As expected, some differences were observed among the medium to poor models and groups. Overall, the top six predicting groups ranked using the fully automated MaxSub are also the top six groups ranked at CASP3. We conclude that MaxSub is a suitable method for the automatic evaluation of models. (+info)
MOTIVATION: Rapid software prototyping can significantly reduce development times in the field of computational molecular biology and molecular modeling. Biochemical Algorithms Library (BALL) is an application framework in C++ that has been specifically designed for this purpose. RESULTS: BALL provides an extensive set of data structures as well as classes for molecular mechanics, advanced solvation methods, comparison and analysis of protein structures, file import/export, and visualization. BALL has been carefully designed to be robust, easy to use, and open to extensions. Especially its extensibility which results from an object-oriented and generic programming approach distinguishes it from other software packages. BALL is well suited to serve as a public repository for reliable data structures and algorithms. We show in an example that the implementation of complex methods is greatly simplified when using the data structures and functionality provided by BALL. (+info)
First pass annotation of promoters on human chromosome 22.
(13/288)
The publication of the first almost complete sequence of a human chromosome (chromosome 22) is a major milestone in human genomics. Together with the sequence, an excellent annotation of genes was published which certainly will serve as an information resource for numerous future projects. We noted that the annotation did not cover regulatory regions; in particular, no promoter annotation has been provided. Here we present an analysis of the complete published chromosome 22 sequence for promoters. A recent breakthrough in specific in silico prediction of promoter regions enabled us to attempt large-scale prediction of promoter regions on chromosome 22. Scanning of sequence databases revealed only 20 experimentally verified promoters, of which 10 were correctly predicted by our approach. Nearly 40% of our 465 predicted promoter regions are supported by the currently available gene annotation. Promoter finding also provides a biologically meaningful method for "chromosomal scaffolding", by which long genomic sequences can be divided into segments starting with a gene. As one example, the combination of promoter region prediction with exon/intron structure predictions greatly enhances the specificity of de novo gene finding. The present study demonstrates that it is possible to identify promoters in silico on the chromosomal level with sufficient reliability for experimental planning and indicates that a wealth of information about regulatory regions can be extracted from current large-scale (megabase) sequencing projects. Results are available on-line at http://genomatix.gsf.de/chr22/. (+info)
Validating clustering for gene expression data.
(14/288)
MOTIVATION: Many clustering algorithms have been proposed for the analysis of gene expression data, but little guidance is available to help choose among them. We provide a systematic framework for assessing the results of clustering algorithms. Clustering algorithms attempt to partition the genes into groups exhibiting similar patterns of variation in expression level. Our methodology is to apply a clustering algorithm to the data from all but one experimental condition. The remaining condition is used to assess the predictive power of the resulting clusters-meaningful clusters should exhibit less variation in the remaining condition than clusters formed by chance. RESULTS: We successfully applied our methodology to compare six clustering algorithms on four gene expression data sets. We found our quantitative measures of cluster quality to be positively correlated with external standards of cluster quality. (+info)
Basecalling with LifeTrace.
(15/288)
A pivotal step in electrophoresis sequencing is the conversion of the raw, continuous chromatogram data into the actual sequence of discrete nucleotides, a process referred to as basecalling. We describe a novel algorithm for basecalling implemented in the program LifeTrace. Like Phred, currently the most widely used basecalling software program, LifeTrace takes processed trace data as input. It was designed to be tolerant to variable peak spacing by means of an improved peak-detection algorithm that emphasizes local chromatogram information over global properties. LifeTrace is shown to generate high-quality basecalls and reliable quality scores. It proved particularly effective when applied to MegaBACE capillary sequencing machines. In a benchmark test of 8372 dye-primer MegaBACE chromatograms, LifeTrace generated 17% fewer substitution errors, 16% fewer insertion/deletion errors, and 2.4% more aligned bases to the finished sequence than did Phred. For two sets totaling 6624 dye-terminator chromatograms, the performance improvement was 15% fewer substitution errors, 10% fewer insertion/deletion errors, and 2.1% more aligned bases. The processing time required by LifeTrace is comparable to that of Phred. The predicted quality scores were in line with observed quality scores, permitting direct use for quality clipping and in silico single nucleotide polymorphism (SNP) detection. Furthermore, we introduce a new type of quality score associated with every basecall: the gap-quality. It estimates the probability of a deletion error between the current and the following basecall. This additional quality score improves detection of single basepair deletions when used for locating potential basecalling errors during the alignment. We also describe a new protocol for benchmarking that we believe better discerns basecaller performance differences than methods previously published. (+info)
Gene structure prediction and alternative splicing analysis using genomically aligned ESTs.
(16/288)
With the availability of a nearly complete sequence of the human genome, aligning expressed sequence tags (EST) to the genomic sequence has become a practical and powerful strategy for gene prediction. Elucidating gene structure is a complex problem requiring the identification of splice junctions, gene boundaries, and alternative splicing variants. We have developed a software tool, Transcript Assembly Program (TAP), to delineate gene structures using genomically aligned EST sequences. TAP assembles the joint gene structure of the entire genomic region from individual splice junction pairs, using a novel algorithm that uses the EST-encoded connectivity and redundancy information to sort out the complex alternative splicing patterns. A method called polyadenylation site scan (PASS) has been developed to detect poly-A sites in the genome. TAP uses these predictions to identify gene boundaries by segmenting the joint gene structure at polyadenylated terminal exons. Reconstructing 1007 known transcripts, TAP scored a sensitivity (Sn) of 60% and a specificity (Sp) of 92% at the exon level. The gene boundary identification process was found to be accurate 78% of the time. also reports alternative splicing patterns in EST alignments. An analysis of alternative splicing in 1124 genic regions suggested that more than half of human genes undergo alternative splicing. Surprisingly, we saw an absolute majority of the detected alternative splicing events affect the coding region. Furthermore, the evolutionary conservation of alternative splicing between human and mouse was analyzed using an EST-based approach. (See http://stl.wustl.edu/~zkan/TAP/) (+info)