expression_profiling - trinityrnaseq/BerlinTrinityWorkshop2018 GitHub Wiki

Extract transcript clusters by expression profile by cutting the dendrogram

Let's revisit the edgeR_trans/ directory and further explore the patterns of differentially expressed transcripts.

% cd ~/workspace/edgeR_trans

And take another look at the heatmap of clustered transcript expression profiles from within your browser (view file: "diffExpr.P1e-3_C2.matrix.log2.centered.dat.genes_vs_samples_heatmap.pdf")

The transcript expression profile clustering on the left side of the heatmap defines several large clusters of transcripts that have similar expression patterns across the different samples.

Extract clusters of transcripts with similar expression profiles by cutting the transcript cluster dendrogram at a given percent of its height (ex. 60%), like so:

# Cut the tree into clusters:
% $TRINITY_HOME/Analysis/DifferentialExpression/define_clusters_by_cutting_tree.pl \
       --Ptree 60 -R diffExpr.P1e-3_C2.matrix.RData

This creates a directory containing the individual transcript clusters, including a pdf file that summarizes expression values for each cluster according to individual charts. From within your browser, visit the new subdirectory "diffExpr.P1e-3_C2.matrix.RData.clusters_fixed_P_60" and view the file "my_cluster_plots.pdf".