Project Roadmap

[schae234]

Road Map

cob is an open source software package used to visualize gene co-expression networks created with Camoco. Using COB, users are able to browse and interact with "Camoco datasets" using a GUI either on their own machines are through the web.

The original COB tool and functionality is described in this paper.

Motivation and Background

Camoco is a tool used to integrate data from gene co-expression networks with data from genome-wide association studies (GWAS). To better understand what COB does, its important to understand the output from Camoco, which involved both GWAS and co-expression data. We will briefly cover them below, but for a more in depth review, see this review and the original Camoco paper.

In species with low quality or modest gene annotations, co-expression networks can be a powerful tool to predict what biological processes under-annotated genes are involved in. Briefly, gene co-expression networks quantify a genes expression levels across many (20+) experiments (called its profile) and compares them to the profile of all the other genes in the genome. Genes with highly similar profiles are typically biologically related. This is extremely useful for under-annotated genes. For example, if unknown gene X had a highly similar gen expression profile (i.e. strongly co-expressed) with many of the genes known to be involved in photosynthesis, its likely that at some level, gene X is also involved at some level in photosynthesis. Often co-expression analysis involved building networks from gene expression data and looking at co-expression of sets of genes with previously described biological function (e.g. GO Terms, KEGG). In short, using guilt by association, researchers can quickly predict what genes do based on their gene expression patterns in species that lack high quality annotations.

Often, researchers are not necessarily searching for what a gene does, but instead, what genes are responsible for a specific trait of interest. For example, if you were interested in what genes improved the nutritional quality in corn, you might set up a genetic study that attempts the map the genes involved in this process. One popular approach is a genome-wide association study or GWAS which looks for associations between mutations (called SNPS) and traits. Gene annotations, or lack thereof, play a huge part here since the output of a GWAS are SNPs that act as markers near causal genes, so the resolution of a GWAS is limited.

Camoco takes as input a gene expression matrix, a GFF file designating gene locations in the genome, and a set of SNPs associated with a trait. Camoco's "overlap" algorithm looks for sets of genes near GWAS SNPs that exhibit higher than expected co-expression. The final output of this "overlap" analysis is a TSV table to candidate genes near your GWAS SNPs as well as the co-expression scores determined by the algorithm.

Using COB to visualize Camoco datasets

This gene list produced by the overlap algorithm essentially ranks your list of candidate genes (near GWAS SNPs) based on co-expression, however visualizing the relationships between the genes in the network are not easily when the data are in table form. Enter COB.

COB is a web based GUI interface to Camoco datasets, it is used to browse and visualize the genes networks. The tools itself relies on datasets built using Camoco, and only provides visualization to the output of the Camoco. Currently COB is great for basic, topical browsing Camoco datasets, but lacks advanced functionality and other niceties offered by a graphical interface. The rest of this document outlines the road map and trajectory of the project.

Milestones

Short Term Goals - what we are working on right now

• Documentation and tutorials
• UX reflow and webpage layout tweaks
• Adding in other pre-built Camoco images / information

Medium Term Goals - what we want to have next

• Build Camoco datasets with COB
• Browse high level networks without GWAS data
• Browse Overlap output

Long Term Goals - Our future vision for COB

• Run overlap analyses from COB
• Browsing network levels (zoom feature)
• Speedup and optimzations