Documente Academic
Documente Profesional
Documente Cultură
5/11/12 11:41 AM
Home
Documentation
Analysis
Query
MAGIA DOCUMENTATION
MAGIA (MiRNA And Genes Integrated Analysis web tool) allows the user to access a target prediction database (Query option) and to carry out a miRNA and genes expression profiles integrated analysis, by adopting different statistical measures of profiles relatedness, algorithms for expression profiles combination and target prediction methods. This tutorial comprises two sections, with examples of settings and results, for both the "Query" and the "Analysis" pipelines.
OUTLINE
QUERY - Browse the target prediction database QUERY SECTION DESCRIPTION EXAMPLE 1: Selection of target predictions for three different miRNAs ANALYSIS - Integrative analysis of target predictions and miRNAs/genes expression profiles. INTRODUCTION TO THE ANALYSIS SECTION: matched and not-matched expression matrices EXAMPLE 2: Integrative analysis WITH MATCHED expression matrices EXAMPLE 3: Integrative analysis WITH NOT MATCHED expression matrices EXAMPLE 4: HOW TO USE Cytoscape for miRNA-network visualisation from MAGIA results
http://gencomp.bio.unipd.it/magia/documentation/
Page 1 of 17
Magia Documentation
5/11/12 11:41 AM
Example 1: Results
For the selected miRNA, the complete list of predicted targets is given as list of predicted relationships, each hyperlinked to different databases
http://gencomp.bio.unipd.it/magia/documentation/
Page 2 of 17
Magia Documentation
5/11/12 11:41 AM
For all selected miRNA, the complete list of predicted target is also given as a downloadable text file.
mirna
gene/transcript miranda pita 57462 3360 56886 55964 11163 6497 6645 11016 651 1641 569.0 460.0 1251.0 487.0 558.0 553.0 785.0 493.0 613.0 707.0 -14.34 -11.28 -15.78 -12.22 -12.04 -11.16 -13.87 -18.28 -12.4 -15.81
hsa-let-7e hsa-let-7e hsa-let-7e hsa-let-7e hsa-let-7e hsa-let-7e hsa-let-7e hsa-let-7e hsa-let-7e hsa-let-7e ...
Expression matrices must be tab-delimited text files; the first row must contain sample names; the first column must contain gene/miRNA IDs Matched data: sample names are sample IDs and must be exactly the same in both matrices! In the figure below, a schema is shown describing how to prepare the matrices in the MATCHED SAMPLES case.
http://gencomp.bio.unipd.it/magia/documentation/
Page 3 of 17
Magia Documentation
5/11/12 11:41 AM
Measures
MAGIA provides different measures and methods for the analysis of matched data. Spearman Correlation: non parametric, rank-based linear correlation measure, suitable for non-normally distributed data and/or small sample size (e.g. 3 to 5). Pearson Correlation: parametric linear correlation measure, suggested for normally distributed data and medium-large sample size (>5). Mutual Information: a classic information measure quantifying the mutual dependence of variables, including non-linear relationships. Suitable for large sample size (>20 needed). Notice that both highly positively and negatively correlated vectors are associated to high mutual information. Genmir: Combined analysis based on a Variational Bayesian method. Suitable only for sparse incidence matrices of target predictions. The most intuitive and simple to interpret measures are those based on correlations. Obviously, the power of the analysis depends on the number of available samples (technical replicates do not contribute much information) and on the variability of expression profiles in considered samples. This is the reason why we recommend a preprocessing of expression matrices to eliminate miRNAs and/or genes almost invariable. NOT MATCHED SAMPLES When matched data are not available one option is to collect miRNA and genes expression data produced on similar samples. For instance, miRNAs only expression data in interesting normal and tumor samples have been produced in a lab and gene expression data in the same type of biological samples (but not the same) are available in public database. The data collection and processing will give rise to two matrices, for miRNAs and genes, say with 30 and 40 samples respectively. Each matrix include two classes of samples (T for tumor and N for normal), continuing our example we may have, 15 T plus 15 N for the miRNA matrix and 10 T plus 30 N for the gene matrix. The meta analysis separately considers each matrix to identify expression profiles significantly variable among classes (which may be two or more) and integrate results with target predictions. The power of the meta-analysis of not-matched data is generally less that of the integrated analysis of matches data, also depending on sample size and real similarity among considered samples, but can indeed give interesting indications for hypothesis generation and experimental design.
Expression matrices
Expression matrices must be tab-delimited text files; the first row must contain sample names; the first column must contain gene/miRNA IDs Unmatched data: the sample name represent the sample class and the same classes should be present in both matrix samples In the figure below, a schema is shown describing how to prepare the matrices in the NOT MATCHED SAMPLES case.
http://gencomp.bio.unipd.it/magia/documentation/
Page 4 of 17
Magia Documentation
5/11/12 11:41 AM
http://gencomp.bio.unipd.it/magia/documentation/
Page 5 of 17
Magia Documentation
5/11/12 11:41 AM
http://gencomp.bio.unipd.it/magia/documentation/
Page 6 of 17
Magia Documentation
5/11/12 11:41 AM
Sample miRNA and gene expression matrices, fully compliant with user-selected settings, are downloadable at the Step 3 of the Analysis (Tip: download sample expression matrices, GSE14834 and use them for the analysis) The user is also allowed to select a subset of rows IDs to be considered for the integrated analysis (optional, leave blank to consider all IDs in the matrix) Example 2: STEP 4 - RESULTS
SUMMARY RESULT PAGE - This page shows the network and the table of top top 250 regulatory relationships supported by expression data analysis and gives links to the files containing all relationships and allowing functional enrichment analysis.
http://gencomp.bio.unipd.it/magia/documentation/
Page 7 of 17
Magia Documentation
5/11/12 11:41 AM
Gene-centered page - For a selected gene, this page shows all miRNAs resulting to be included regulatory relationships supported by expression data analysis. For each gene and miRNA hyperlinks to different databases are provided.
http://gencomp.bio.unipd.it/magia/documentation/
Page 8 of 17
Magia Documentation
5/11/12 11:41 AM
miRNA-centered page - For a selected miRNA, this page shows all genes resulting to be included regulatory relationships supported by expression data analysis. For each gene and miRNA hyperlinks to different databases are provided.
Functional enrichment analysis - Top supported target genes of miRNAs (default 250), are directly uploaded, with corresponding settings on the DAVID page, for functional enrichment analysis (mapping to pathways and knowledge maps, ...) This is done by exploiting David APIs (http://david.abcc.ncifcrf.gov/content.jsp?file=DAVID_API.html), imposing a maximum of 400 genes and of 2048 URL characters. When the user selection is not compliant with these limits warning messages suggest to reduce the number of genes or to conduct directly David analysis on gene list extracted from the .TSV file.
http://gencomp.bio.unipd.it/magia/documentation/
Page 9 of 17
Magia Documentation
5/11/12 11:41 AM
Cytoscape file - A text file is also provided, to be imported in Cytoscape: software for network visualization and analysis
Integrative analysis WITH NOT MATCHED expression matrices Example 3: STEP 1 - Method
Select ID type (RefSeq and ENSEMBL transcripts, EntrezGene and ENSEMBL genes) and "meta analysis" from the list of available methods for the integrated analysis. This means that, separately for genes/transcripts and miRNAs in available sample classes, MAGIA will calculate LIMMA p-values of differential expression, which are then combined by using the inverse chi square distribution to identify oppositely variable miRNA-gene pairs.
http://gencomp.bio.unipd.it/magia/documentation/
Page 10 of 17
Magia Documentation
5/11/12 11:41 AM
http://gencomp.bio.unipd.it/magia/documentation/
Page 11 of 17
Magia Documentation
5/11/12 11:41 AM
Sample miRNA and gene expression matrices, fully compliant with user-selected settings, are downloadable at the Step 3 of the Analysis (Tip: download sample expression matrices and use them for the analysis) The user is also allowed to select a subset of rows IDs to be considered for the integrated analysis (optional, leave blank to consider all IDs in the matrix) Example 3: STEP 4 - RESULTS
SUMMARY RESULT PAGE - This page shows the network and the table of top top 250 regulatory relationships supported by expression data analysis and gives links to the files containing all relationships and allowing functional enrichment analysis.
http://gencomp.bio.unipd.it/magia/documentation/
Page 12 of 17
Magia Documentation
5/11/12 11:41 AM
Gene-centered page - For a selected gene, this page shows all miRNAs resulting to be included regulatory relationships supported by expression data analysis. For each gene and miRNA hyperlinks to different databases are provided.
miRNA-centered page - For a selected miRNA, this page shows all genes resulting to be included regulatory relationships supported by expression data analysis. For each gene and miRNA hyperlinks to different databases are provided.
http://gencomp.bio.unipd.it/magia/documentation/
Page 13 of 17
Magia Documentation
5/11/12 11:41 AM
Functional enrichment analysis - For the selected set of genes induced by the ranked interaction (default 250), data are directly uploaded, with corresponding settings on the DAVID page, for functional enrichment analysis page (mapping to pathways and knowledge maps, ...)
Cytoscape file - A text file is also provided, to be imported in Cytoscape: software for network visualization and analysis
HOW TO USE Cytoscape for miRNA-network visualisation from MAGIA results Example 4: STEP 1 - Relationships are given from MAGIA
Download the .tsv file from MAGIA
http://gencomp.bio.unipd.it/magia/documentation/
Page 14 of 17
Magia Documentation
5/11/12 11:41 AM
http://gencomp.bio.unipd.it/magia/documentation/
Page 15 of 17
Magia Documentation
5/11/12 11:41 AM
http://gencomp.bio.unipd.it/magia/documentation/
Page 16 of 17
Magia Documentation
5/11/12 11:41 AM
http://gencomp.bio.unipd.it/magia/documentation/
Page 17 of 17