Catálogo de publicaciones - libros

Understanding biochemical pathways is one of the big challenges of the field of molecular biology nowadays. Computer science can contribute in this area in a variety of ways. One of them is providing formalisms and tools to simulate and check properties of pathways. One formalism that is well known and suited for modeling concurrent and distributed systems is Milner’s Calculus of Communicating Systems (CCS). CCS is a process algebra and there are many tools that support modeling and automatic verification of properties of systems modeled in terms of CCS processes. This paper describes the regulation of the lactose operon using CCS. We validate our formal model by automatic checking a series of properties that are known for the regulation of the lactose. Thus, we show the viability of using process algebras to model and reason about biochemical networks.

One of the most important challenges for Bioinformatics is the simulation of a single cell, even if we restrict ourselves to simple models of the molecular networks responsible for the behavior of organisms. The challenge involves not only the development of experimental techniques to obtain kinetic parameters that characterize the myriad reactions occurring inside cells, but also computational approaches able to simulate and test the complex models generated. These systems have stochastic behavior; they can take different paths depending on environmental conditions. We can describe them using stochastic models that have a high computational cost, but the simulations can be performed efficiently on distributed architectures like grids and clusters of computers. In this work we describe an implementation of a computational architecture to execute this kind of large scale simulation using a grid infrastructure. We validate the proposed architecture using experiments in order to estimate its performance.

The enzymes of the shikimate pathway constitute an excellent target for the design of new antibacterial agents. This pathway is found in bacteria, fungi, plants and apicomplexan parasites but is absent in mammals. Chorismate Synthase (CS) catalyzes the last step of this pathway, the product of which is utilized in other enzymatic transformations like the biosynthesis of aromatic amino acids, folate, vitamin K and ubiquinone. This reaction is the most unusual of the entire pathway and is unique in nature. It converts EPSP to chorismate in the presence of a reduced FMN cofactor. Structure prediction used the comparative protein structure modeling methodology. The three-dimensional (3D) structure prediction of the enzyme was performed using the crystal structure (PDB ID: 1QX0) of CS from as template (≈42% identity), and the MODELLER6v2 package. Additionally, in order to understand the possible binding modes of substrate and cofactor to the enzyme EPSP and FMN, respectively, were geometrically docked to CS. FMN binding to CS of (MTB) is similar to that of the template despite the change of Asn251 in to Gln256 in MTB. The longer side chain of Gln256 is overlapping with the FMN cofactor and a small conformational change is needed in order to properly accommodate this interaction. EPSP binding mode is also very similar to that of the template with three hydrogen bonds missing. This could be due to artifacts from the simple geometric docking we performed. Refinement with energy-based docking algorithms should relax the enzyme and substrates, thus promoting the expected interactions between them. Understanding the structure of MTB CS together with its cofactor and substrate binding modes should facilitate the search for inhibitors of this enzyme as alternative agents to treat tuberculosis.

Secondary structure prediction methods are widely used bioinformatics algorithms providing initial insights about protein structure from sequence information. Significant efforts to improve the prediction accuracy over the past years were made, specially the incorporation of information from multiple sequence alignments. This motivated the search for the factors contributing for this improvement. We show that in two of the highly ranked secondary structure prediction methods, DSC and PREDATOR, the use of multiple alignments consistently improves the prediction accuracy as compared to the use of single sequences. This is validated by using different measures of accuracy, which also permit to identify that helical regions benefit the most from alignments, whereas -strands seem to have reached a plateau in terms of predictability. Also, the origins of this improvement is explored in terms of sequence specificity, secondary structure composition and the extent of sequence similarity which provides the optimal performance. It is found that divergent sequences, in the identity range of 25–55% provide the largest accuracy gain and that above 65% identity there is almost no advantage in using multiple alignments.

BLAST homology searches have been largely used to annotate function to novel sequences. Secondary databases like KOG can be used in this intention since their sequences have functional classification. We devised an experiment where public ESTs from four eukariotic organisms, which protein sequences are present in the KOG database, are classified to functional KOG categories using tBLASTn. First we assigned the ESTs from one organism to KTL (KOG, TWOG and LSEs) proteins and then we searched the database depleted of the same organism’s proteins to simulate a novel transcriptome. Data show that classification was correct (assignment equals annotation) 87.2%, 96.8%, 92.0%, 88.7% for (Ath), (Cel), (Dme) and (Hsa) respectively. We have estimated identity cutoffs for all organisms to use with tBLASTn. These cutoffs trim the same amount of events that a BLASTn in order to minimize false positives in consequence of sequence errors. We found values of 80%, 78%, 78% and 84% for amino-acid identity cutoff for Hsa, Dme, Cel and Ath, respectively. We then evaluated our system by comparing the KTL categories of the assigned ESTs with the KTL categories that the ESTs were classified without the organism’s KTL proteins. Moreover, we show the potential of annotation of the KOG database and the ESTs used. Suplementary Information can be found at: http://www.biodados.icb.ufmg.br

Whether diet has been influencing the genomic and proteomic constitution of the organisms along the evolution is an interesting and not answered question. Here, we investigate the hypothesis that essential amino acids – the ones that are not produced by the organisms – have being replaced in proteins by non-essential ones. We compare the amino acid composition of the proteome from human, worm and fly, that cannot synthesize all amino acids, with the ones from plant, baker yeast and budding yeast, capable to synthesize all of them. The analysis was made with 190,074 proteins composed of 87,175,891 amino acids. Our data seems to evidence a little bias on the usage of non-essential amino acids by the metazoan organisms, except for the worm. Thus, the preliminary results shown here support the thesis that non-essential ones have replaced essential amino acids.

The large amount of data from complete genomes gave rise to the need for computational tools to analyze and compare them. In this work, we propose a method for comparing three genomes simultaneously, at the basic level of their sequences. This comparison can indicate the set of genes shared by genomes, giving interesting clues about the metabolic pathways and proteins related to particular issues. The input for the method is three sets of gene coding sequences or products and the output are the sequences exclusive to each genome, the sequences common to pairs of genomes, and the sequences common to the three genomes. Because each sequence in a genome may be similar to many sequences in the other two genomes, some complicated situations may arise. The main feature of our method is the ability to avoid such situations. We used our method to compare genomes of two pathogenic and five non-pathogenic fungi, and made a biological analysis based on one of these results.

Aligning cDNA sequences to genomic sequences is a very common way to study expressed sequences, find their genes, and study alternative splicing. Several computer programs address this problem, using heuristics to define exon regions. Usually, standard alignment algorithms are not used to align ESTs to genomic DNA, due to the existence of large regions of introns. This paper compares the EST-to-genomic alignments produced by , , and standard sequence aligners using an appropriate score. Surprisingly, standard aligners performed quite well with sequences having few errors.

The objective of this paper is to describe a new approach for locating the centromere of each chromosome displayed in the digitalized photomicrography of fish cells. To detect the centromere position, the authors propose methods for both image segmentation and split touching chromosomes based on the fuzzy sets theory and a method for the rotation of chromosomes. These methods were applied to two species of fish chromosomes: and . Using a database with 40 images including metacentric, submetacentric and subtelocentric chromosomes, and comparing the centromere locating obtained by the proposed algorithm with the manual results obtained by two expert cytogeneticists, the average accuracies were 81.79% and 82.54% respectively.

Efficient family classification of newly discovered protein sequences is a central problem in bioinformatics. We present a new algorithm, using , which identifies equivalences between the amino acids in different positions of a motif for each family. We also show that better classification can be achieved identifying representative fingerprints in the amino acid chains.