inst/shiny/markdown/info.md

In kkmann/pwcuratr: Functional Pathway Cluster Curation tool

Candidate Gene Pathway Clusters

tl;dr: elicit candidate gene networks based on seed-genes.

This tool allows the defintion of candidate gene networks based on a set of seed genes, a combination of reactome pathway information and predicted functional gene-gene interactions following

Wu, G., Feng, X., & Stein, L. (2010). A human functional protein interaction network and its application to cancer data analysis. Genome biology, 11(5), R53.

This version uses the 2018 release of the function interaction dataset directly downloaded from https://reactome.org/download-data. Internally all mappings are done via an ENSEMBL version 97 snapshot downloaded using the biomaRt package.

The app is part of the R package pwcuratr.

Define Seed-Genes

The first tab allows three different options for specifying the set of seed genes.

1. If a curated pathway cluster already exists and has been downloaded from the app before, the .zip file can be uploaded and all settings are restored to continue working.
2. A set of genes can be uploaded directly as .csv file. The file must contain a column 'ensembl_gene_id' with ENSEMBL gene (or transcript) identifiers. Version suffixes are not supported since they are not consitently used in the underlying databases. All genes will be added to the current pool of potential seed genes and automatically be selected.
3. Individual genes can be added to the current pool of seed genes via the dropdown selector. The selector support fuzzy search over gene names and ENSEMBL gene identifiers when text is entered directly.

Select Pathways

This tab allows to match the selected set of seed genes to reactome.org biological pathway identifiers. The process must be started manually by clicking on the button 'update from seed gene selection'. All reactome pathways in which at least one of the seed genes participates will be listed with clickable links to the respective reactome.org page. After revising the set of seed genes, the button must be clicked again but all currently selected pathways are retained if they contain at least one of the revised seed genes. If pathway cluster was restored by uploading a saved .zip file, the previously selected pathways will be restored when clicking 'update from seed gene selection'.

The pathways are sorted by number of participating genes (ascending) to prioritize specific (small) pathways.

Pruning & Plotting

To further increase the specificity of the functional interaction neighborhood defined by the pathway cluster, the resulting set of candidate genes (all genes contained in the selected reactome pathways plus the original seed genes) can be pruned. Top this end, we use a recent (2018) release of a set of predicted functional gene-gene interactions using methodology for Wu et al, 2010. Every interaction is scored between 0 (low confidence) and 1 (high confidence). The set of interactions used to define a functional neighborhood is controlled by specifying the 'minimal score' input. Finally, the parameter $k$ controls the maximal number of edges between any of the seed genes and a candidate gene. I.e., all candidate genes whose closest connect to any of the seed genes is greater than $k$ will be pruned from the set.

To guide the selection of the two pruning parameters, the number of retained genes and the number of connected components in the final graph are plotted against $k$ with the current selection highlighted in green.

The individual connected components of the graph can be selected for plotting via a dropdown menu on the left and the dimensions of the plot can be tailored manually to the number of genes.

To save the current curated pathway cluster, simply click the download button. The .zip file can then be used to restore the same pathway cluster by uploading the file on the tab 'Define Seed Genes'.

kkmann/pwcuratr documentation built on March 7, 2020, 2 a.m.