Tailor-made solution for extracting some plots from a specific dataset.

This tool contains a fully connected graph of genes in the data files, as well as presence absence data of those genes in an annotated set of genomes. The tool will ask you for a focal gene around which to build a network, along with thresholds to filter graph edges. It will compute the local subgraph.

Follow the instructions at install_instructions.txt

You will be asked to provide several parameters:

• Gene name. Must exactly match a gene that exists in the network. Using the interactive mode, suggestions will be made if your input doesn't match.
• p threshold. Edges have p-values between 0 and 1. The p threshold specifies the upper bound of p value and will remove all edges with p > this bound.
• LR threshold. Edges have numeric LR values. The LR theshold specifies the lower bound of LR values and will remove edges with LR < the threshold.
• d. Must be an integer. The number of neighbors away from the focal node. Providing a value of 0 will cause the program to return a graph containing only edges incident on the focal node.

From the project directory, simply run python extract_heatmaps.py. An interactive terminal will guide you through the process.

If you wish to quickly run several graphs, you may provide your parameters in a csv file. The file format is highly inflexible. It must use a comma as the delmiter. It must contain exactly four columns; gene, p, lr, d, in that order. Each row in the file will correspond to a new run. The parameters of each row will be validated individually. Console messages will indicate whether a row was successful or not.

To use csv mode, run python extract_heatmap.py path/to/input.csv. An example is available in the data folder.

For each gene and set of parameters, the program will create a directory with several results files. The program will create a results directory if one doesn't exist. It will also create subdirectories named for the focal genes of each run. Within the gene-named directories, the program will create a directory named for the parameters used. For example, if a user runs gene=vanA, p=0.05, lr=100, d=1, the program will create: results/vanA/p0.05_lr100.0_d1/

Within the result directory, the program will save 4 files:

• gene_habitat_dist_proportion.png. This file shows the number of genomes of each habitat containing the focal gene. On the y axis, the proportion of total genomes which the gene appears in is shown. e.g. (n genomes of habitat Y / n genomes gene appears in).
• gene_habitat_dist.png. Similar to above, but y axis shows genome counts
• gene_PA_heatmap.png. A clustergram showing the presence/absence pattern of all genes in the computed subgraph
• gene_neighborhood_graph.graphml. A graphml file containing the computed subgraph. May be loaded into cytoscape.

The program will generate a .graphml file which may be loaded into Cytoscape. We have also prepared a Cytoscape Style File, data/NicheGraph.xml. To use this, load the .graphml file into Cytoscape using file -> import -> network from file. Then, import the style file using file -> import -> style from file. You may then select "Niche Style" from the styles tab.