SpectConnect for metabolomics
When analyzing GC/MS data from multiple conditions, metabolomics researchers typically compare the experimental spectra with those stored in a reference library. But some peaks cannot be easily matched to spectra of known metabolites. To bypass the identification step, Gregory Stephanopoulos and colleagues at the Massachusetts Institute of Technology developed SpectConnect. This algorithm tracks unidentified metabolites that are present in multiple samples by comparing all of the spectra for every sample. In addition, the algorithm determines which metabolites differentiate the samples. SpectConnect relies on the increase in S/N observed when many replicates are run, so several technical and biological replicates are a requirement for optimum performance.
Stephanopoulos and colleagues tested the algorithm on standard mixtures and on the metabolomes of three E. coli strains. Almost all of the known compounds in the standards were detected; the only exception was that isoleucine and leucine could not be distinguished with their methods. When the researchers analyzed the E. coli samples with SpectConnect and conventional methods, the new algorithm detected more potential biomarkers. (Anal. Chem. 2007, DOI: 10.1021/ac0614846)
ProteomeCommons.org IO Framework
Developed by Phil Andrews and colleagues at the University of Michigan, the ProteomeCommons.org Input and Output (IO) Framework helps researchers to analyze MS data. For example, the IO Framework can extract raw data from various file formats, including proprietary ones. With this feature, the data can be easily converted to other formats, such as mzData and mzXML. In addition, peptide and protein sequence files can be manipulated. Also, the tool can shuffle or reverse protein sequences so users can test for false positives. Another feature of the IO Framework is that it filters mass spectral data on the basis of intensity or m/z range. The tool is open-source and freely available at www.proteomecommons.org/current/531. (Bioinformatics 2007, DOI: 10.1093/bioinformatics/btl573)
tYNA for comparative interactomics
Several systems exist that allow researchers to visualize networks of genes or proteins, but so far, none of them provide complex analysis capabilities. Therefore, Mark Gerstein and co-workers at Yale University and the Dana-Farber Cancer Institute developed TopNet-like Yale Network Analyzer (tYNA), an interactive web-based system that compares many networks simultaneously. Although the system already includes the most widely used network data sets, it also can accept new data in several popular formats. Statistics, such as the clustering coefficient and betweenness, are calculated by tYNA. To identify hubs and bottlenecks in networks, users can filter the data on the basis of a statistical cutoff value. In addition, the system identifies various motifs, such as chains, cycles, and feed-forward loops. tYNA can highlight deficiencies in some networks, including defective cliques in which edges may be missing. Analyses can be performed on several networks at once; these networks then can be merged into one composite network. The results can be downloaded and saved in several formats. (Bioinformatics 2006, 22, 2968–2970)
IntNetDB for protein-protein interactions
To provide the scientific community with an up-to-date database of reliable protein–protein interactions (PPIs), Jing-Dong Han and co-workers at the Chinese Academy of Sciences have developed IntNetDB. The database is a compilation of PPIs generated by seven analytical and bioinformatics methods. A likelihood ratio is calculated to represent the probability that a PPI as generated by each method is correct. The likelihood ratios are integrated by the Naïve Bayes model into one score. The PPI network derived from these analyses can be viewed with an online visualization tool called intView. A tool to extract clusters is also available. Han and co-workers plan to update the database regularly as new PPI information is published. The web interface is freely available by contacting Han. (BMC Bioinformatics 2006, 7, 508)
Return to Top
Copyright © American Chemical Society, 1155 Sixteenth Street N.W., Washington, DC 20036
