Using this, and the experimental design factors from the Array Express and GEO databases, DvD identifies a main factor for the experiment. Red edges are for inverse correlations and green positiveĭvD expects as input either a drug or disease profile. ( B) Example visualization produced by the Cytoscape plug-in for the prostate cancer profile (gse17906). This function calculates and identifies significant Enrichment scores and produces corresponding network files. Classifyprofile can take input from generateProfiles, selectrankedlists or the users own preprocessed data. SelectRankedLists can be used to select a subset of the contrasts output from generate profiles. Finally, differential expression statistics are calculated using limma. Probes to Genes maps Affymetrix probes to HUGO gene symbols using BiomaRt. GenerateProfiles imports the data and normalizes CEL files where necessary. ( A) GenerateProfiles and ClassifyProfile are wrapper functions whose stages are shown in the vertical flow charts. Multiple probes mapping to the same gene can be converted using the average or maximal intensities, median polish or by selecting the probe with the highest variance across all arrays.ĭvD pipeline. Probes mapping to multiple gene identifiers are removed. Annotation files can be passed to DvD to process data from other platforms. DvD will automatically annotate, filter and combine probes to HUGO genes for the Affymetrix platforms HG-U133A, HG-U133A-2 and HG-U133-Plus2 using BiomaRt ( Durinck et al., 2009). Data are normalized using either rma or mas5 ( Irizarry et al., 2003). Options to import data from local directories and Array Express or GEO are supported ( Davis et al., 2007 Kauffmann et al., 2009). The DvD pipeline provides a number of processing options to generate genome-wide expression profiles from microarray experiments (see Fig. In the final networks, drug or disease nodes are linked to the DrugBank ( and Medical Subject Headings (MeSH) ( web browsers, respectively. The Cytoscape ( plug-in provides a user interface to the full DvD pipeline, as well as a visualization platform for the results. DvD is flexible, offering customizable and default data options for the input profiles and the reference data. This reference data has associated networks where, unlike similar tools, drugs or diseases exerting similar effects on transcription are grouped into clusters. With DvD, users can automatically compare input profiles to reference drug data from CMap and disease profiles curated from GEO. DvD differs from existing web servers and databases (such as ProfileChaser, MARQ and SPIED, see Supplementary Material) in that it can dynamically access both Array Express and GEO to generate input profiles. This is done by making use of gene set enrichment analysis ( Subramanian et al., 2005) and visualizing the final results in networks containing clusters of similar drugs and diseases. Motivated by this, we have developed Drug versus Disease (DvD), an R package to ‘match’ drug and disease profiles. It is expected that analysing new and existing data from public repositories such as Array Express ( Gene Expression Omnibus (GEO) ( and the Connectivity Map (CMap) ( using these methods will become increasingly popular in computational drug discovery ( Iorio et al., 2012). This highlights the potential power of mining existing safe compounds for repurposing, which does not require the expensive and extensive initial design and clinical phases of drug discovery. These methods have resulted in hypotheses of differential uses (repurposing) for existing compounds that have been validated experimentally. The central paradigm is that a drug compound, which shows the opposite effect on gene expression to the observed for a disease could be used to treat that particular disease. Multiple methods based on matching gene expression signatures have been proposed to identify anti-correlated drug and disease profiles ( Hu et al., 2009 Sirota et al., 2011).
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |