sampDis {chemodiv}

Calculate sample dissimilarities

Description

Function to calculate dissimilarities between samples. Either Bray-Curtis dissimilarities and/or Generalized UniFrac dissimilarities are calculated.

Usage

sampDis(sampleData, compDisMat = NULL, type = "BrayCurtis", alpha = 1)

Arguments

sampleData

Data frame with the relative concentration of each compound (column) in every sample (row).

compDisMat

Compound dissimilarity matrix, as calculated by compDis. If this is supplied, Generalized UniFrac dissimilarities can be calculated.

type

Type of sample dissimilarities to be calculated. This is Bray-Curtis dissimilarities, type = "BrayCurtis", and/or Generalized UniFrac dissimilarities, type = "GenUniFrac".

alpha

Parameter used in calculations of Generalized UniFrac dissimilarities. alpha can be set between 0 and 1. With alpha = 0, equal weight is put on every branch in the dendrogram. With alpha = 1, branches are weighted by their abundance, and hence more emphasis is put on high abundance branches. alpha = 0.5 strikes a balance between the two. alpha 0.5 or 1 is recommended, with alpha = 1 as default. See Chen et al. 2012 for details.

Details

The function calculates a dissimilarity matrix for all the samples in sampleData, for the given dissimilarity index/indices. Bray-Curtis dissimilarities are calculated using only the sampleData. This is the most commonly calculated dissimilarity index used for phytochemical data (other types of dissimilarities are easily calculated using the vegdist function in the vegan package).

If a compound dissimilarity matrix, compDisMat, is supplied, Generalized UniFrac dissimilarities can be calculated, which also use the compound dissimilarity matrix for the sample dissimilarity calculations. For the calculation of Generalized UniFrac dissimilarities (Chen et al. 2012), the compound dissimilarity matrix is transformed into a dendrogram using hierarchical clustering (with the UPGMA method). Calculations of UniFrac dissimilarities quantifies the fraction of the total branch length of the dendrogram that leads to compounds present in either sample, but not both. The (weighted) Generalized UniFrac dissimilarities implemented here additionally take compound abundances into account. In this way, both the relative proportions of compounds and the biosynthetic/structural dissimilarities of the compounds are accounted for in the calculations of sample dissimilarities, such that two samples containing more biosynthetically/structurally different compounds have a higher pairwise dissimilarity than two samples containing more biosynthetically/structurally similar compounds. As with Bray-Curtis dissimilarities, Generalized UniFrac dissimilarities range in value from 0 to 1.

Value

List with sample dissimilarity matrices. A list is always outputted, even if only one matrix is calculated.

References

Bray JR, Curtis JT. 1957. An Ordination of the Upland Forest Communities of Southern Wisconsin. Ecological Monographs 27: 325-349.

Chen J, Bittinger K, Charlson ES, et al. 2012. Associating microbiome composition with environmental covariates using generalized UniFrac distances. Bioinformatics 28: 2106-2113.

Lozupone C, Knight R. 2005. UniFrac: a New Phylogenetic Method for Comparing Microbial Communities. Applied and Environmental Microbiology 71: 8228-8235.

Examples

data(minimalSampData)
data(minimalCompDis)
sampDis(minimalSampData)
sampDis(sampleData = minimalSampData, compDisMat = minimalCompDis,
type = c("BrayCurtis", "GenUniFrac"), alpha = 0.5)

data(alpinaSampData)
data(alpinaCompDis)
sampDis(sampleData = alpinaSampData, compDisMat = alpinaCompDis,
type = "GenUniFrac")