Canvas Browser I
LINCS Canvas Browser I and II (LCBI and LCBII), developed at the Ma’ayan Lab, are web-based tools that enable users to explore thousands of genome-wide gene expression experiments performed on six breast cancer cell lines (BT-20, MCF7, SK-BR-3, MCF 10A, Hs 578T and MDA-MB-231). The browser visualizes results from L1000 experiments where drugs or endogenous ligands were applied to six human breast cancer cell lines in different concentrations and expression measured at different time points. The visualization of the results is organized by cell line and batch where perturbations that induced similar responses are clustered together on a canvas.
Clicking on a specific experiment on the canvas of experiments, displayed on the left, results in enrichment analyses displayed on various canvases on the right. The canvases on the right represent gene set libraries and the enriched terms are gene sets that highly overlap with the genes that were up- or down-regulated in the selected experiment on the left.
There was a problem loading that gene set library. Please choose a different library.
On the left are the L1000 experiments (red canvas). This canvas is a visualization of data collected from experiments in one cell line from one batch. Each square on the canvas represents one condition (each condition was repeated 4 times and the average behavior is retained). The experiments are organized by their expression vector similarity computed using Pearson’s correlation coefficient applied to 1000 genes measured by the L1000 technology. Simulated annealing is applied to find an arrangement with good, but not necessarily maximal fitness. The canvas has no axis, and the edges “fold in” on themselves forming a torus. The initial view of the canvas shows the significance of the responses in each experimental condition (perturbation) as compared to the control. This is the overall average change when comparing the control expression vector to the treated condition expression vector, where the brighter the square the more significant the change. In other words, for dark spots, the ligand or drug did not have an observable effect. The slider below the canvas controls the overall brightness of the grid, and the cell line and dataset selectors determine which data is shown. Above the canvas are links to other views that color experiments on the canvas by time point, drug/ligand, or dose. The location of experiments on the canvas is fixed, so when toggling between views each square of the canvas continues to represent the same condition/perturbation.
On the right are canvases representing gene-set libraries (blue canvas). Each square in these canvases is a gene-set associated with a specific functional term such as a pathway, gene ontology term, or transcription factor. These canvases are also arranged to cluster gene lists with similar content near each other. Bright spots represent areas on the canvas where the lists are highly similar. There are several gene-set libraries to choose from, and the default is KEGG pathways. The gene-set libraries are divided into various broad categories which are: transcription, pathways, ontologies, drugs/diseases and cell types.
Clicking a square on the experiments canvas triggers a gene-list enrichment analysis using the currently displayed gene-set library and either the up- or down-regulated genes in that experiment. The display switches to the “p-value” view where the top 20 enriched terms are highlighted while the rest of the terms are colored black. The most highly significant terms (those terms with the highest overlap) are colored with the brightest color representing the lowest p-values. You can mouse over the squares on the canvas to see the labels of the enriched terms. In addition, the results can be viewed as a table or a network. The p-values are computed using the Fisher exact test. Using the mouse wheel you can zoom the canvases in and out as well as pan by clicking and dragging.
Version 2 of the Canvas Browser is also available. It links out to the Ma'ayan Lab's Enrichr tool to perform gene set analyses.