Da-Fu Ding

Horizontal gene transfer (HGT), a process through which genomes acquire genetic materials from distantly related organisms, is believed to be one of the major forces in prokaryotic genome evolution. However, systematic investigation is still scarce to clarify two basic issues about HGT: (1) what types of genes are transferred; and (2) what influence HGT events over the organization and evolution of biological pathways. Genome-scale investigations of these two issues will advance the systematical understanding of HGT in the context of prokaryotic genome evolution. Having investigated 82 genomes, we constructed an HGT database across broad evolutionary timescales. We identified four function categories containing a high proportion of horizontally transferred genes: cell envelope, energy metabolism, regulatory functions, and transport/binding proteins. Such biased function distribution indicates that HGT is not completely random; instead, it is under high selective pressure, required by function restraints in organisms. Furthermore, we mapped the transferred genes onto the connectivity structure map of organism-specific pathways listed in Kyoto Encyclopedia of Genes and Genomes (KEGG). Our results suggest that recruitment of transferred genes into pathways is also selectively constrained because of the tuned interaction between original pathway members. Pathway organization structures still conserve well through evolution even with the recruitment of horizontally transferred genes. Interestingly, in pathways whose organization were significantly affected by HGT events, the operon-like arrangement of transferred genes was found to be prevalent. Such results suggest that operon plays an essential and directional role in the integration of alien genes into pathways.

Two-component and phosphorelay signal transduction systems are central components in the virulence and antimicrobial resistance responses of a number

Summary: In the genomic era, researchers often want to know more information about a biological sequence by retrieving its related articles. However, there is no available tool yet to achieve conveniently this goal. Here we developed a new literature-mining tool MedBlast, which uses natural language processing techniques, to retrieve the related articles of a given sequence. An online server of this program is also provided. Availability: Both online server and the program are available freely at http://medblast.sibsnet.org

High throughput scoring algorithms that are used to find the match of a tandem mass spectrum to a predicted mass spectrum of a peptide within a database have been applied in shotgun proteomics. However, these algorithms could produce a significant number of incorrect peptide identifications. Here a novel approach was developed to scoring tandem mass spectra against a peptide database, in which fragment ion probabilities, number of enzymatic termini of candidate peptides, matching quality and match pattern between experimental and theoretical spectrum were considered. Benchmarking the novel scorer on a large set of experimental MS/MS spectra, it is demonstrated that PepSearch performs significantly better than the widely used software SEQUEST. The PepSearch software is available at http://compbio.sibsnet.org/projects/pepsearch.

Glial cell line-derived neurotrophic factor (GDNF) has received much attention as potential therapeutic agent for the treatment of neurodegenerative diseases. It will be very important to discover the molecular mechanism of this factor and its specific GFRalpha co-receptor. Based on the principle of molecular evolution that site-specific functional importance is relevant to the pressure it undergoes under natural selection, evolutionary trace method was used to identify the functional epitopes in GDNF and GFRalpha families. Some trace residues had been proved to be important in ligand-receptor binding, especially in rat GFRalpha1, where alanine scanning mutagenesis confirmed that sites N(152)N(153), R(259), S(316)N(317)S(318) and Q(247)D(248)S(249) were critical for GFRalpha1 binding to GDNF or Ret and thus affected the formation of GDNF-GFRalpha1-Ret complex.

We present a simple method to detect pathogenicity islands and anomalous gene clusters in bacterial genomes. The method uses iterative discriminant analysis to define genomic regions that deviate most from the rest of the genome in three compositional criteria: G+C content, dinucleotide frequency and codon usage. Using this method, we identify many virulence-related gene islands, e.g. encoding protein secretion systems, adhesins, toxins, and other anomalous gene clusters, such as prophages. The program and the whole dataset, including the catalogs of genes in the detected anomalous segments, are publicly available at http://compbio.sibsnet.org/projects/pai-ida/. This program can be used in searching for virulence-related factors in newly sequenced bacterial genomes.

Furin/kexin processing proteases catalyze the proteolysis of large protein precursors involved in many biological processes, such as zymogen activation, peptide hormone synthesis, viral protein processing and receptor maturation, making them potential targets for therapeutic agents. Herein, homology modeling and weighted evolutionary tracing were combined to investigate the interactionmechanism of furin/kex2 with eglin C mutants. The model structures showed that there were many acidic residues in the furin (kex2) binding interface, contributing to specificity for multiple basic residues of their corresponding substrates or inhibitors. Besides, some rational explanations were presented for the different inhibitor/substrate specificity of the furin/kexin members by combining the model structures with results of evolutionary tracing. Based on these analyses,an attempt was made to rationally redesign the eglin C by interface engineering with heterogeneous self-consistent ensemble optimization to improve its inhibitory specificity on furin/kex2. With the model complex structures of furin/kex2 and eglin C variants as structural templates, the P(1), P(2) and P(4) of eglin C were redesigned, respectively. The design results show that both furin and kex2 favored basic residues at P(1), P(2) and P(4) in eglin C, in good agreement with the experimental data. The detection of many specific residues in S' part of furin/kexin sequences made possible designing inhibitors with high specific binding to furin and kex2, respectively. As for furin, the best inhibitor designed was eglin C-P(2)'Glu-P(3)'Asp-P(4)'Arg (only these three positions were shown), while the best eglin C variant for kex2 designed was P(2)'Arg-P(3)'Arg-P(4)'Glu. The structures show that furin and kex2 form distinct interactions with these two eglin C variants. Herein, a strategy was proposed that combine homology modeling, evolutionary tracing and rational interface redesign to investigate enzyme-inhibitor interactions and inhibitor engineering. This computational design gives some rational guidance to further experimental inhibitor engineering.

High-density cDNA microarray is important in monitoring large scalegenomic expression profile. With radiolabeled cDNA sample, nylon filter cDNA microarrays have high sensitivity and wide linear range of hybridization signals, so it can be employed to detect many important low-abundance cDNAs. However, the nylon filter tends to be randomly contaminated, and the array spots are prone to dispersion and saturation, resulting in inaccurate expression value. Based on aphysical model, we address a novel treatment with accurate positioning, complement of saturation and correction of interference. Thus the accurate quantificati on of expression can be extracted with remarkably improved reproducibility.

Peptide sequencing via tandem mass spectrometry(MS/MS)is one of the most powerful tools in proteomics to identify proteins. A new algorithm was developed for de novo interpretation of MS/MS spectra using graph theory and dynamic alignment between real spectra and theoretical spectra. The trustworthy peptides from de novo interpretation were used in protein identification via database searching. A high throughput statistical analysis of SwissProt and TrEMBL protein databases showed that it's enough to identify a protein in database with three sequence tags of four amino acid residues, two sequence tags of five amino acid residues or one sequence tag of eight amino acid residues.