Proceedings. International Conference on Intelligent Systems for Molecular Biology最新文献

A data base of minimally frustrated alpha helical segments is defined by filtering a set comprising 822 non redundant proteins, which contain 4783 alpha helical structures. The data base definition is performed using a neural network-based alpha helix predictor, whose outputs are rated according to an entropy criterion. A comparison with the presently available experimental results indicates that a subset of the data base contains the initiation sites of protein folding experimentally detected and also protein fragments which fold into stable isolated alpha helices. This suggests the usage of the data base (and/or of the predictor) to highlight patterns which govern the stability of alpha helices in proteins and the helical behavior of isolated protein fragments.

I present in this work an algorithm for deriving protein functional annotations which are position-specific. The input is based on the results of a sequence similarity search of the query sequence against a sequence database. Strings of words are extracted from the descriptions of the proteins, and the correlation between proteins having the same descriptors and the amino acid conservation is used to compute a score that indicates which descriptor is likely to describe better the function of each particular residue. Analysis of the score curves and comparison of different functions allows an easy detection of parts of the sequence associated to different function. Different levels of functional specificity can be compared, allowing to choose the one that suits better the function of the protein. Immediate applications of this algorithm are, support for (automated) methods of protein functional annotation, and database coherence check.

This is an investigation of methods for finding short motifs that only occur in a fraction of the input sequences. Unlike local search techniques that may not reach a global optimum, the method proposed here is guaranteed to produce the motifs with greatest z-scores. This method is illustrated for the Ribosome Binding Site Problem, which is to identify the short mRNA 5' untranslated sequence that is recognized by the ribosome during initiation of protein synthesis. Experiments were performed to solve this problem for each of fourteen sequenced prokaryotes, by applying the method to the full complement of genes from each. One of the interesting results of this experimentation is evidence that the recognized sequence of the thermophilic archaea A. fulgidus, M. jannaschii, M. thermoautotrophicum, and P. horikoshii may be somewhat different than the well known Shine-Dalgarno sequence.

One current approach to quality control in DNA array manufacturing is to synthesize a small set of test probes that detect variation in the manufacturing process. These fidelity probes consist of identical copies of the same probe, but they are deliberately manufactured using different steps of the manufacturing process. A known target is hybridized to these probes, and those hybridization results are indicative of the quality of the manufacturing process. It is not only desirable to detect variations, but also to analyze the variations that occur, indicating in what process step the manufacture changed. We describe a combinatorial approach which constructs a small set of fidelity probes that not only detect variations, but also point out the manufacturing step in which a variation has occurred. This algorithm is currently being used in mass-production of DNA arrays at Affyetrix.

Correlations between sequence separation (in residues) and distance (in Angstrom) of any pair of amino acids in polypeptide chains are investigated. For each sequence separation we define a distance threshold. For pairs of amino acids where the distance between C alpha atoms is smaller than the threshold, a characteristic sequence (logo) motif, is found. The motifs change as the sequence separation increases: for small separations they consist of one peak located in between the two residues, then additional peaks at these residues appear, and finally the center peak smears out for very large separations. We also find correlations between the residues in the center of the motif. This and other statistical analysis are used to design neural networks with enhanced performance compared to earlier work. Importantly, the statistical analysis explains why neural networks perform better than simple statistical data-driven approaches such as pair probability density functions. The statistical results also explain characteristics of the network performance for increasing sequence separation. The improvement of the new network design is significant in the sequence separation range 10-30 residues. Finally, we find that the performance curve for increasing sequence separation is directly correlated to the corresponding information content. A WWW server, distanceP, is available at http://www.cbs.dtu.dk/services/distanceP/.

Screening for potential ligands and docking them into the binding sites of proteins is one of the main tasks in computer-aided drug design. Despite the progress in computational power, it remains infeasible to model all the factors involved in molecular recognition, especially when screening databases of more than 100,000 compounds. While ligand flexibility is considered in most approaches, the model of the binding site is rather simplistic, with neither solvation nor induced complementary usually taken into consideration. We present results for screening different databases for HIV-1 protease ligands with our tool Slide, and investigate the extent to which binding-site conformation, solvation, and template representation generate bias. The results suggest a strategy for selecting the optimal binding-site conformation, for cases in which more than one independent structure is available, and selecting a representation of that binding site that yields reproducible results and the identification of known ligands.

Optical mapping is a novel technique for generating the restriction map of a DNA molecule by observing many single, partially digested, copies of it, using fluorescence microscopy. The real-life problem is complicated by numerous factors: false positive and false negative cut observations, inaccurate location measurements, unknown orientations and faulty molecules. We present an algorithm for solving the real-life problem. The algorithm combines continuous optimization and combinatorial algorithms, applied to a non-uniform discretization of the data. We present encouraging results on real experimental data.

The application of image registration techniques resulted in the construction of an integrated atlas of Drosophila segmentation gene expression in both space and time. The registration method was based on a quadratic spline approximation with flexible knots. A classifier for automatic attribution of an embryo to one of the temporal classes according to its gene expression pattern was developed.)

We have used the Teiresias algorithm to carry out unsupervised pattern discovery in a database containing the unaligned ORFs from the 17 publicly available complete archaeal and bacterial genomes and build a 1D dictionary of motifs. These motifs which we refer to as seqlets account for and cover 97.88% of this genomic input at the level of amino acid positions. Each of the seqlets in this 1D dictionary was located among the sequences in Release 38.0 of the Protein Data Bank and the structural fragments corresponding to each seqlet's instances were identified and aligned in three dimensions: those of the seqlets that resulted in RMSD errors below a pre-selected threshold of 2.5 Angstroms were entered in a 3D dictionary of structurally conserved seqlets. These two dictionaries can be thought of as cross-indices that facilitate the tackling of tasks such as automated functional annotation of genomic sequences, local homology identification, local structure characterization, comparative genomics, etc.

Methods based on the Mutual Information statistic (MI methods) predict structure by looking for statistical correlations between sequence positions in a set of aligned sequences. Although MI methods are often quite effective, these methods ignore the underlying phylogenetic relationships of the sequences they analyze. Thus, they cannot distinguish between correlations due to structural interactions, and spurious correlations resulting from phylogenetic history. In this paper, we introduce a method analogous to MI that incorporates phylogenetic information. We show that this method accurately recovers the structures of well-known RNA molecules. We also demonstrate, with both real and simulated data, that this phylogenetically-based method outperforms standard MI methods, and improves the ability to distinguish interacting from non-interacting positions in RNA. This method is flexible, and may be applied to the prediction of protein structure given the appropriate evolutionary model. Because this method incorporates phylogenetic data, it also has the potential to be improved with the addition of more accurate phylogenetic information, although we show that even approximate phylogenies are helpful.