LINCS Workflow

Find novel compounds that mimic or reverse a disease signature

The LINCS Center for Transcriptomics collected gene expression data from human cells treated with thousands of chemical and biological perturbagens using the L1000 technology. This large dataset was processed into signatures: lists of up and down genes, or vectors of differentially expressed genes. This resource of molecular gene expression signatures can be used to find perturbagens that mimic or reverse your own gene expression signature, or any external signature you might be interested in studying.

For example, a disease gene expression signature can be defined as the differentially regulated genes that characterize a disease by comparing normal tissue to diseased tissue. With LINCS data you can search for compounds that can either reverse or mimic the disease signature. The top ranked compounds may have therapeutic potential. There are currently two web-apps that can be used to perform this type of query.

The CLUE Query App

The CLUE website has a query tool that enables users to find perturbagens (e.g. compounds or RNAi knockouts) that mimic or reverse their input signature. The query tool can be accessed here after logging in.


Query App

The CLUE Query App takes lists of up- and down-regulated genes as input. The results page shows the top compounds, knockdowns, and overexpression perturbations that match your signature input.


clue.io Results Page

L1000 Characteristic Direction Signature Search Engine (L1000CDS2)

The BD2K-LINCS DCIC has developed an alternative L1000 query tool. This tool enables users to search only a subset of the L1000 small molecules data. The tool's underlying database computes signatures differently. It uses the Characteristic Direction (CD) method to select significant perturbations as well as prioritize differentially expressed genes. The CD processed L1000 signatures are served through a state-of-the-art web-based search engine application called L1000CDS2(http://amp.pharm.mssm.edu/L1000CDS2). The L1000CDS2 search engine provides prioritization of thousands of small molecule signatures, and their pairwise combinations, predicted to either mimic or reverse an input gene expression signature using two methods. The L1000CDS2 search engine also predicts drug targets for all the small molecules profiled by the L1000 assay that were processed. Targets are predicted by computing the cosine similarity between the L1000 small molecules signatures and a large collection of signatures extracted from the gene expression omnibus (GEO) for single gene perturbations in mammalian cells.

The L1000CDS2 homepage is shown below. A user can input their disease gene signatures as a weighted list of genes (not shown) or separate up and down genes (shown on the left).


L1000CDS2 Query Page

The default is to search for perturbagens that reverse the user’s signature. Clicking the search button directs the user to their results page.


L1000CDS2 Results Page

The results page shows the user the top perturbagens that reverse their input disease signature. The user can also identify which genes are affected by the perturbation using the overlap button or by viewing a clustergram of their results, shown below:


L1000CDS2 Clustergram

The clustergram view shows the user’s input disease signature as rows, perturbagens as columns and the effect of a perturbagen on a gene’s expression: red/blue squares indicate increased/decreased expression.

L1000CDS2 enables users to easily identify perturbagens that mimic or reverse their disease signature of interest using LINCS L1000 data. For more information please view the help documentation at http://amp.pharm.mssm.edu/L1000CDS2/help/.

Generating disease signatures from GEO

The BD2K-LINCS DCIC developed a tool that allows users to create their own disease signatures. Using GEO2Enrichr, (http://amp.pharm.mssm.edu/g2e) and GEO (Gene Expression Omnibus http://www.ncbi.nlm.nih.gov/geo), a user can find gene expression data from a disease of interest, compute a signature and then submit it to the CLUE Query app or to L1000CDS2 for drugs/small-molecules.

Start by following the installation instructions of GEO2Enrichr and install the browser extension as shown below for Chrome.


G2E on the Google Play Store

After installing the browser extension, a user can perform a search for a disease on GEO and identify a study of interest as shown below for Huntington’s disease KO mouse model:


Selecting Samples with G2E

After selecting the control and experimental samples and pressing the GEO2Enrichr button you will be shown a submission page.


G2E Submission Form

Here you can add meta-data and choose your analysis parameters (e.g. differential expression method) and submit your data for analysis. When the analysis has finished you can open your results tab (shown below) to download your disease signature or perform more analyses.


G2E Results Page

You can now download the up and down genes and submit them to the CLUE Query app or simply click on the L1000CDS2 icon to open the results from the L1000CDS2 analysis.