Library of Integrated Network-Based Cellular Signatures (LINCS)
Data now available : To access computational and experimental results from Columbia University’s LINCS U01 centers, go to geWorkbench , the Department of Systems Biology’s desktop application for integrated genomics. Types of data that are now available include:
- drug-drug synergy results from titration experiments
- computed drug mechanism of action results
- drug-drug similarity results obtained by calculating the distance between the mechanisms of action of two drugs
geWorkbench offers a number of visualizations, including titration curves, heat maps, and network displays of drug mechanisms of action.
Please note that the collection of data that is now available is preliminary and ongoing. The final collection of data is expected soon.
The Library of Integrated Network-based Cellular Signatures (LINCS) is a multi-institutional, collaborative project initiated by the National Institutes of Health whose goal is to identify and categorize molecular signatures that occur when cells are exposed to agents that perturb their normal function. Understanding how perturbations affect the behavior of cells is important because it can provide insights into the causes of disease and help researchers develop new therapeutic strategies.
LINCS approaches this problem using tools from systems biology, chemical biology, computational biology, and other disciplines, including both high-throughput experimentation and sophisticated mathematical analysis. LINCS centers are using high-throughput screening to observe how various cell lines respond to molecules such as siRNAs and bioactive compounds, and are developing computational methods for analyzing and categorizing the signatures that result. The project is also developing an open access database and web-based tools that will foster a wide range of biological and biomedical research, enabling the broader scientific community to identify molecules within biological networks that hold promise as targets for new disease therapies.
Systematically perturbing cells with pairs of agents will enable us to better understand the mechanisms of action of many drugs.
The Columbia University Department of Systems Biology was selected as a member of the LINCS consortium in 2011. We currently have two grants, the first focusing on technology development, and a second on the design of new computational tools. By using high-throughput experimentation to perturb cells with pairs of agents in a very systematic fashion, our goal is to generate and categorize the molecular signatures that result in a way that will enable us to better understand the mechanisms of action (MoA) of many drugs. Using a variety of computational algorithms we also aim to predict — without the time and expense of actually testing every possible combination in the laboratory — which combinations of small molecules are most likely to have synergistic effects in live cells.
The principal investigator for Columbia’s LINCS projects is Andrea Califano.
In addition to generating and analyzing molecular perturbation data, the Department of Systems Biology has also developed software that offers the scientific community access to the large repository of experimentally derived and computationally inferred drug synergy and mechanism of action data we are producing. Users can query the results of computations and experiments and probe for predicted and validated synergistic effects of drug pairs on a number of different cancer cell lines. The software also provides a graphical user interface to enable advanced visualization and interaction with the data. Additionally, we are developing an application programming interface (API) that will make LINCS data usable in third-party applications. The new software is included in geWorkbench , the bioinformatics platform of the National Center for the Multiscale Analysis of Genomic and Cellular Networks (MAGNet), also based at Columbia University.
LINCS Project 1:
Profiling Signatures of Synergistic Chemical Perturbations in Diverse Cellular Contexts
In the past, understanding complex drug interactions was largely a trial and error exercise. Researchers in the Department of Systems Biology are working to make combination therapy a rational discipline. By analyzing LINCS data and other publicly available datasets, we are generating molecular profile information that will improve scientists’ abilities to predict how drug combinations will interact and to identify combination therapies that may improve treatment for diseases such as cancer. This pilot study will:
- generate a set of experimental assays — primarily using high-throughput, high-content methods — that can be used to characterize phenotypic changes in specific cell types
- use a systems biology approach to generate a prioritizied list of likely synergistic drug combinations, based on existing expression profiles and IC50 profiles from LINCS and Connectivity Map (CMap) data
- test drug combinations that our methods predict to be synergistic and compare experimental results to those of random drug combinations. This effort will establish objective benchmarks for estimating the performance of predictive algorithms.
- characterize the synergistic behavior of drugs with orthogonal mechanism of action vs. drugs sharing a similar MoA. (Orthogonal mechanisms of action are MoAs that share little or no overlap.)
While our main goal is to identify drug synergies that will improve the treatment of disease, this project will also enable us to generate several new functional interaction maps of cellular phenotypes (interactomes), including disease phenotypes such as breast carcinoma.
Accessing drug synergy results
geWorkbench offers a simple interface for building custom queries based on the results of synergy experiments conducted at Columbia University. The annotations available for narrowing a query include the tissue type, cell line, drug names, assay type, and synergy measurement type. The results are presented in a tabular view, with visualization options including:
- heat map display of drug vs. drug synergy score results
- network view using Cytoscape to depict synergy strength between each pair of drugs
- titration curves depicting the underlying data used to calculate synergy for individual drug pairs
LINCS Project 2:
A Systems Approach to Elucidate Mechanisms of Drug Activity and Sensitivity Using LINCS Data
Researchers in the Department of Systems Biology have developed a repertoire of experimentally validated algorithms such as ARACNe (for the dissection of transcriptional interactions) and MINDy (for post-translational interactions in mammalian cells). We have also pioneered the development of methods for integrating information from various layers within complex regulatory networks. These include the algorithms MARINa (used to interrogate interactomes and to identify master regulator and master integrator genes that cause specific phenotypes), and IDEA (used to elucidate compound mechanism of action).
Building on this methodological base, Columbia has begun to introduce novel approaches and develop new algorithms that are specifically designed for the analysis of LINCS data. As part of this project, we are currently working to:
- elucidate the MoA and activity of small molecule and RNAi/cDNA mediated perturbations
- identify genes that mediate a cell’s sensitivity or resistance to the activity of specific compounds
- identify gene-gene, gene-compound, and compound-compound interactions whose modulation can synergistically induce a desired phenotype (e.g., apoptosis, cell cycle arrest, or loss of pluripotency)
- use LINCS in vitro signatures to predict compound-related properties such as MoA, activity, sensitivity, synergy, and other phenomena in vivo
Accessing drug activity and sensitivity data
geWorkbench offers a simple interface for building custom queries based on the results of drug similarity experiments conducted at Columbia University. The annotations available for narrowing a query include the tissue type, cell line, drug names, and the similarity algorithm used. As with the synergy query, the results are presented in tabular form, with additional visualizations including:
- heat map display of drug-drug similarity scores
- functional mode of action (fMoA) of a drug as represented by regulatory gene differential activity in response to the drug as determined using VIPER
- network view of the differential expression of targets of selected regulatory genes using Cytoscape. This is the data underlying the VIPER calculation.