Sisana

SiSaNA now has the option to create clustermaps, allowing the user to visualize clusters of samples and genes/TFs. For this option, users also specify categorical metadata columns to color the samples on. For example, if you have columns called "Tumor_Grade" and "Group" in your metadata, you can create something like this (shown is just the top of the clustermap):


To create these clustermaps, the following can be added to the params.yml file:

Use the following to create a clustermap, which will cluster the parameters (TFs, genes, etc.) and add column colors for categorization of samples
example: sisana visualize clustermap params.yml
clustermap:
datafile: Path to file containing the expression or indegrees of each gene per sample
filetype: Type of input file, must be either "csv", "txt", or "tsv", where csv implies comma separated values and txt/tsv implies tab-separated
statsfile: Path to file with the comparison output performed between two groups
metadata: Path to the csv metadata file mapping samples to groups (groups must match names of the groupnames arg), must have a header of the format 'name,group'
genelist: .txt file containing a list of genes to plot, must match the name of genes in the datafile. Recommended to not use more than 10 genes, otherwise use a heatmap.
column_cluster: Flag ("True" or "False") for if you wish to perform hierarchical clustering on the columns. If False, then the samples will only be grouped according to the first category in category_label_columns
row_cluster: Flag ("True" or "False") for if you wish to perform hierarchical clustering on the genes
plot_gene_names: Flag ("True" or "False") for whether to plot the names of the genes on the rows of the heatmap
plot_sample_names: Flag ("True" or "False") for whether to plot the names of the samples on the columns of the heatmap
category_label_columns: The names of columns (categories of samples) in the metadata file you wish to use for coloring the columns in the clustermap. The clustermap will be grouped by the first name given, unless column_cluster=True
- Tumor_Grade
- Group
category_column_colors: The matplotlib color codes you wish to use for the sub-categories given in category_label_columns. For example, a category called "Sex" may want to use ["pink", "blue"] as their colors (though SiSaNA has no way of determining which samples are female or male, so you may need to change the order you give the colors)
- ["r", "b", "green", "y", "teal"]
- ["dodgerblue", "orange", "k", "lightcoral", "thistle"]
outdir: Path to directory to output file to
prefix: Prefix to use for the output file; note that the output file will automatically be generated with the suffix '_clustermap.png'

With this newest version of SiSaNA, all analysis is now performed with the use of a params.yml file instead of specifying arguments directly via the command line.

Many features are now available in the new version. These include the following:

1. Filtering of all PPI, motif, and gene expression files, which is a prerequisite for running PANDA/LIONESS.
2. Filtering the output of Lioness for only edges found in the prior.
3. Calculation of the in-and out-degree of genes and TFs, respectively.
4. Reducing the number of decimal places in either the PANDA/LIONESS output or the calculated in-/out-degrees, which greatly saves on storage space.
5. Extraction of specific TFs/genes, which is useful for analyses such as limma (see Ritchie et al., 2015). An option to perform analyses with limma is not available as part of this software.
6. Comparison of groups identified in dimension reduction techniques such as UMAP or tSNE. These include comparisons of TFs/genes between two groups, survival analysis between groups, and gene set enrichment analysis (GSEA)
7. Visualization of these results via volcano plots, box plots, violin plots, and heatmaps is also possible.

Initial release, can perform the lioness pipeline, including calculating in- and out-degree and comparing degrees between groups.