phylosmith

phylosmith is a supplementary package to build on the phyloseq-objecy from the phyloseq package. Phyloseq-objects are a great data-standard for microbiome and gene-expression data, this package is aimed to provied easy data-wrangling and visualization.

A lot of these functions are just to make “data-wrangling” easier for investigators. Others will implement complex routines in a, hopefully, efficient and concise manner. I have also made functions to make figures for quick examination of data, but they may or may not be suitable for publication, as some may require parameter optimization.

Installation

Requirements

Linux Systems

For some Linux systems you may need to install the following two programs through your terminal.

Ubuntu example:

sudo apt-get install libmysqlclient-dev libgdal-dev libudunits2-dev

These programs are required by some dependancies and may not come in your default OS distribution.

Windows Systems

if you are working on WINDOWS you likely need to install the CRAN program Rtools.When prompted, select add rtools to system PATH.

R

phylosmith depends on the usage of the phyloseq package released by Dr. Paul McMurdie. The pacakge is maintained on BioConductor, and can be installed through R using the following commands:

if(!requireNamespace("BiocManager", quietly = TRUE)){
  install.packages("BiocManager")
} 
BiocManager::install("phyloseq")

Additionally, the package imports a number of other packages to use their advanced functions. These packages may install with the phylosmith installation, but it is always best to install independently.

install.packages(c("devtools", RcppEigen", "RcppParallel", "Rtsne", "ggforce", "units"))

phylosmith

The package is hosted on Github, and can be installed through R with:

devtools::install_github('schuyler-smith/phylosmith')
library(phylosmith)

Functions

Wrangling

Call	Description
conglomerate_samples	combines samples based on common factor within sample_data
conglomerate_taxa	combines taxa that have same classification
melt_phyloseq	melts a phyloseq object into a data.table
merge_treatments	combines multiple columns in meta-data into a new column
set_sample_order	sets the order of the samples of a phyloseq object
set_treatment_levels	sets the order of the factors in a sample_data column
taxa_extract	creates a phyloseq-object containing only taxa of choice
taxa_filter	filter taxa by proportion of samples seen in
taxa_prune	remove taxa from a phyloseq-object

Analytics

Normalization

Call	Description
library_size	transform abundance data using library-size normalization
relative_abundance	transform abundance data to relative abundance

Taxa

Call	Description
common_taxa	find taxa common to each treatment
taxa_core	filter taxa by proportion of samples and relative abundance
taxa_proportions	computes the proportion of a taxa classification
unique_taxa	find taxa unique to each treatment

Graphics

Abundance

Call	Description
abundance_heatmap	create a ggplot object of the heatmaps of the abundance table
abundance_lines	create a ggplot object of the abundance data as a line graph
phylogeny_profile	create a ggplot barplot object of the compositons of each sample at a taxonomic level
taxa_abundance_bars	create a ggplot object of the abundance of taxa in each sample
taxa_core_graph	create a ggplot object of the core taxa over a range of parameters
variable_correlation_heatmap	create a ggplot heatmatp of the correlation of numerical variables with taxa

Diversity

Call	Description
alpha_diversity_graph	create a ggplot-object box-plot of the alpha-diversity from a phyloseq-object.
dendrogram_phyloseq	create a ggplot-object dendrogram of the distance measurement from a phyloseq-object.
nmds_phyloseq	create a ggplot object of the NMDS from a phyloseq object
pcoa_phyloseq	create a ggplot object of the PCoA from a phyloseq object
tsne_phyloseq	create a ggplot object of the t-SNE from a phyloseq object

Networks

Call	Description
co_occurrence_network	creates a network of the co-occurrence of taxa
network_layout_ps	creates a layout object for a network
network_ps	creates a network object from a phyloseq_object
variable_correlation_network	creates a network of the correlation of taxa and sample variables

Calculations

Call	Description
co_occurrence	calculate co-occurrence between taxa
curate_co_occurrence	filter co-occurrence tables
permute_rho	runs permutations of the otu_table to calculate a significant \(\rho\) value
histogram_permuted_rhos	Create a ggplot object of the distribution of rho values.
quantile_permuted_rhos	calculate quantiles for the permuted rho values from the Spearman-rank co-occurrence
variable_correlation	calculate the correlation of numerical variables with taxa abundances

Datasets

Originally I had created 2 mock phyloseq objects (mock_phyloseq and mock_phyloseq2) that had no real-world data but served to show simple examples of how the functions worked.

Then I decided that I should include a real example of microbiome data (soil_column) becasue it’s always nice to see real examples. soil_column is a published dataset from my lab-group. The data is from an experiment where they looked at the microbial composition of farmland soil before and after manure application, over time, using 16S-sequencing.

Schuyler Smith
Ph.D. Bioinformatics and Computational Biology