Wong AK*, Krishnan A*, Yao V*, Tadych A, Troyanskaya OG. IMP 2.0: a multi-species functional genomics portal for integration, visualization and prediction of protein functions and networks. Nucleic Acids Research. July 2015;43:W128-W133. [PDF] [Method Supplement]
If you use the biological process predictions, please also cite:
Park CY*, Wong AK*, Greene CS*, Rowland J, Guan Y, Bongo LA, Burdine RD, Troyanskaya OG. (2013) Functional Knowledge Transfer for High-accuracy Prediction of Under-studied Biological Processes. PLoS Comput Biol 9(3): e1002957. doi:10.1371/journal.pcbi.1002957i [PDF]
IMP was created by the Laboratory for Bioinformatics and Functional Genomics in the Lewis-Sigler Institute for Integrative Genomics at Princeton University.
Functional genomics can complement traditional sequence similarity to improve the transfer of gene annotations between organisms. In particular, functional knowledge transfer (FKT) can improve the accuracy and coverage of gene-function predictions in under-studied processes (i.e. biological processes that are not well studied in a particular organism of interest but might be well characterized in another organism).
Figure 3 from the FKT PLoS Comput Biol paper shows the potential prediction performance gains.
There are two primary ways to explore the functional networks and biological process predictions in IMP.
Search by gene to view GO (Gene Ontology) biological processes we've predicted your gene to participate in.
Select an organism of interest and enter a gene name (e.g. BRCA1) in the above search box. You will be directed to a gene page with both the predicted processes for your gene of interest and the functional network used in the predictions (centered around your query gene).
Processes that are highlighted are already known to be associated with the query gene.
Search by biological process to view a list of the genes we've predicted to participate in your process of interest.
Select an organism and enter any terms related to your process (e.g. 'dna repair'). The search box will return any GO biological processes matching your query. Select the best matching name and you will be directed to the prediction page for your process.
The displayed network shows functional relationships between a selected gene from the prediction list and genes known to participate in the process, or functional analogs that participate in the process.
The functional networks presented by IMP are highly interactive and configurable. Network views can be controlled by filtering edges by confidence or filtering nodes by connectivity.
Supporting genomic data for functional relationship predictions are readily accessible by hovering over or clicking on any edge in the network. The top ten datasets contributing to the prediction are displayed, and specific evidence from the datasets can be investigated in greater detail.
IMP’s users can create custom gene sets for any organism (for example, a list of genes identified as over-expressed in a microarray experiment) and submit them for analysis. In the analysis, a graphical search of an organism’s gene network is performed on the gene set to retrieve predicted functional neighbors (those likely to participate in the same pathway).
IMP presents these as highly-interactive networks: users can adjust the layout by moving genes, query the evidence supporting a functional relationship, and modify graphical options to customize the display. Importantly, as users adjust network parameters, an enrichment analysis updates in real-time as the network specificity is narrowed or broadened (effectively contracting or expanding the user’s gene set).
Users can compare functional contexts (genes/edges of the network relevant to a biological process) across organisms. The networks are visually aligned using a coordinated layout where movements in the query network are simultaneously updated in the other organisms' networks.