Submit a Revision

*your email :**
page name* : such as "Introduction", "Conclusion"..etc
<strong>2.1 Microarray data</strong><br> Generations of microarray replicates are described in detail<sup></sup>in Tu <em>et al</em>. (<a href="http://bioinformatics.oxfordjournals.org/cgi/content/full/23/13/i282?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=1&andorexacttitle=and&titleabstract=engineering&andorexacttitleabs=and&fulltext=open+access&andorexactfulltext=phrase&search#B22">2002</a>). In brief, mRNA from the Ramos human Burkitt's<sup></sup>lymphoma cell line is used for the experiments. The purified<sup></sup>sample is separated equally into several subgroups and each<sup></sup>subgroup independently goes through the preparation steps. The<sup></sup>final target sample is then divided into several samples and<sup></sup>independently hybridized to 10 different Affymetrix HGU95A arrays.<sup></sup>The data set used for investigating gene co-expression consists<sup></sup>of gene expression profiles from 254 naturally occurring phenotypic<sup></sup>variations of human B-cell. It represents a wide variety of<sup></sup>homogenous B-cell phenotypes derived from normal and tumor related<sup></sup>populations. The microarray experiments are described in Basso<sup></sup><em>et al</em>. (<a href="http://bioinformatics.oxfordjournals.org/cgi/content/full/23/13/i282?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=1&andorexacttitle=and&titleabstract=engineering&andorexacttitleabs=and&fulltext=open+access&andorexactfulltext=phrase&search#B4">2005</a>) and the <em>CEL</em> files are available on the Gene Expression<sup></sup>Omnibus website (series accession number: GSE2350 <a href="http://bioinformatics.oxfordjournals.org/cgi/external_ref?access_num=GSE2350&link_type=NCBIGEO">[NCBI GEO]</a> ).<sup></sup> <p> <a></a><strong>2.2 Permutation of <em>CEL</em> file</strong><br> Raw signal intensities for each probe pairs were randomly permuted<sup></sup>to create uninformative <em>CEL</em> files. We retained the relative<sup></sup>position between PM and MM for every probe pairs, in order to<sup></sup>ensure fair comparison between normalization procedures that<sup></sup>utilize MM information to correct for non-specific binding and<sup></sup>those that rely entirely on PM intensities. However, shuffling<sup></sup>the probe pairs has been sufficient to destroy real signal of<sup></sup>the probe sets as they now consist of random probes values.<sup></sup>This data is crucial in our comparative study as the null set<sup></sup>should not contain any information.<sup></sup> </p> <p> <a></a><strong>2.3 Normalization procedures</strong><br> We compared the four normalization procedures MAS5, RMA, GCRMA<sup></sup>and Li–Wong, and all the normalization were implemented<sup></sup>using software packages available from Bioconductor (<a href="http://www.bioconductor.org/">http://www.bioconductor.org</a>).<sup></sup>We used the default parameters from the software packages unless<sup></sup>otherwise specified. The term ‘Li–Wong’ refers<sup></sup>to the procedure that normalizes arrays using invariant set<sup></sup>of genes and then fits a parametric model to the probe set data,<sup></sup>as described in Li and Wong (<a href="http://bioinformatics.oxfordjournals.org/cgi/content/full/23/13/i282?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=1&andorexacttitle=and&titleabstract=engineering&andorexacttitleabs=and&fulltext=open+access&andorexactfulltext=phrase&search#B14">2001</a>).<sup></sup> </p> <p> <a></a><strong>2.4 Evaluation of biological function relationship</strong><br> GO annotations of the genes were extracted from Affymetrix HGU95<sup></sup>annotation file. There are 10 369 terms for biological process<sup></sup>in total and 61 general terms were removed. We are interested<sup></sup>only in specific terms that are shared by <5% of the genes<sup></sup>in the microarray. A gene pair sharing a common GO term is then<sup></sup>deemed functionally related.<sup></sup> </p> <p> <a></a><strong>2.5 Likelihood ratio of protein–protein interaction</strong><br> We assembled a set of gold-standard positive interactions by<sup></sup>taking the union of interaction data from the Human Protein<sup></sup>Reference Database (HPRD), the Biomolecular Interaction Network<sup></sup>Database (BIND), the Database of Interacting Proteins (DIP)<sup></sup>and IntAct (Bader <em>et al</em>., <a href="http://bioinformatics.oxfordjournals.org/cgi/content/full/23/13/i282?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=1&andorexacttitle=and&titleabstract=engineering&andorexacttitleabs=and&fulltext=open+access&andorexactfulltext=phrase&search#B2">2003</a>; Hermjakob <em>et al</em>., <a href="http://bioinformatics.oxfordjournals.org/cgi/content/full/23/13/i282?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=1&andorexacttitle=and&titleabstract=engineering&andorexacttitleabs=and&fulltext=open+access&andorexactfulltext=phrase&search#B11">2004</a>; Peri<sup></sup><em>et al</em>., 2003; Xenarios <em>et al</em>., <a href="http://bioinformatics.oxfordjournals.org/cgi/content/full/23/13/i282?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=1&andorexacttitle=and&titleabstract=engineering&andorexacttitleabs=and&fulltext=open+access&andorexactfulltext=phrase&search#B25">2002</a>). The resulting gold-standard<sup></sup>positive set consists of 21 509 unique PPIs (heterodimers only)<sup></sup>that could possibly pair up among genes in the Human Genome<sup></sup>U95 array. A negative gold-standard is harder to define, but<sup></sup>we took the common approach by taking the lists of protein pairs<sup></sup>that are unlikely to interact given their cellular localization.<sup></sup>The assembled negative set contains 6 101 360 pairs of proteins<sup></sup>encoded by genes represented on the U95 array. The likelihood<sup></sup>ratio is computed as the fraction of conditional probabilities<sup></sup>for a set of protein pairs, here the top predicted gene pairs<sup></sup>ranked by statistical dependency between expression profiles,<sup></sup>given the gold-standard positive (<em>pos</em>) and negative (<em>neg</em>) sets:<sup></sup> </p> <p> </p> <p> LR = P (coexpressed pair\|<em>pos</em>) / P (coexpressed pair\|<em>neg</em>) </p>
Confirmation code: Enter the code exactly as it appears. All letters are case insensitive, there is no zero.