Diversity calculations¶
Alpha diversity refers to diversity of OTUs/ASVs within a sample. Beta diversity refers to differences in community composition between different sampes. We have three ways of measuring diversity:
- naive diversity considers all OTUs/ASVs to be equally differerent from each (this is also called taxonomic diversity);
- phylogenetic diversity takes evolutionary distances between OTUs/ASVs in a phylogenetic tree into account;
- functional diversity takes pairwise distances between OTUs/ASVs in a distance matrix into account.
In the Hill-based framework used to calculate the naive-, phylogenetic-, and functional diversity measures, the importance of the relative abundance of OTUs/ASVs can be tuned by the diversity order (q).
- A q of 0 means that relative abundance is not considered and OTUs/ASVs are treated as either present or absent in the samples.
- A q of 1 means that that each OTU/ASV is weighted exactly according to its relative abundance in the sample.
- A q > 1 means that more weight is given to OTUs/ASVs with high relative abundance.
Alpha diversity¶
diversity.naive_alpha(tab, q=1)
Returns taxonomic (or naive) alpha diversity values (i.e. Hill numbers) for all samples.
tab is a frequency table holding the read counts, typically obj[‘tab’]
q is the diversity order.
diversity.phyl_alpha(tab, tree, q=1, index='PD'):
Returns alpha diversity values that takes phylogenetic tree distances between sequences into account.
tab is a count table holding the read counts, typically obj[‘tab’]
tree is the tree dataframe, typically obj[‘tree’].
q is the diversity order.
index can be: ‘PD’, ‘D’, or ‘H’. ‘PD’ is the phylogenetic diversity, ‘D’ is the Hill diversity number corrected for phylogenetic relationships, and ‘H’ is the phylogenetic entropy. See Chao et al. (2010). Phil. Trans. R. Soc. B, 365, 3599-3609 (DOI: 10.1098/rstb.2010.0272).
diversity.func_alpha(tab, distmat, q=0, index='FD'):
Returns alpha diversity values that takes the pairwise distances from between sequences into account. The values are the same as the functional diversity measures described by Chiu et al. (2014), PLOS ONE, 9(7), e100014 (DOI: 10.1371/journal.pone.0100014).
tab is a count table holding the read counts, typically obj[‘tab’]
distmat is a dataframe with pairwise distances between the sequences of OTUs/SVs. The file can be generated wit the stats.sequence_comparison() function.
q is the diversity order for the Hill numbers.
index can be: ‘FD’, ‘MD’, or ‘D’. ‘FD’ is the functional diversity (“the effective total distance between species”), ‘MD’ is the mean functional diversity (“the effective sum of pairwise distances between a species and all other species”), and ‘D’ is the functional Hill number (“the effective number of equally abundant and equally distinct species”). See Chiu et al. (2014), PLOS ONE, 9(7), e100014 (DOI: 10.1371/journal.pone.0100014), the descriptions in quotations marks are from that paper.
Beta diversity (dissimilarity)¶
The difference in community composition between different samples can be expressed as beta diversity or dissimilarity. Beta diversity takes a value between 1 (if the compared samples are identical to each) and N (which is the number of samples being compared. For functional beta diversity it goes between 1 and NxN. Dissimilarity takes a value between 0 (if the compared samples are identical) and 1 (if the compared samples are completely different). Dissimilarity can be calculated with a local or regional viewpoint. The local viewpoint quantifies the effective proportion of OTUs/ASVs in a sample that is not shared across all samples. The regional viewpoint quantifies the effective proportion of OTUs/ASVs in the pooled samples that is not shared across all samples. See Chao and Chiu (2016), Methods in Ecology and Evolution, 7, 919-928.
diversity.naive_beta(tab, q=1, dis=True, viewpoint='local'):
Returns a dataframe of pairwise dissimilarities between samples. Taxonomic (or naive) Hill-based dissimilarities of order q are calculated.
tab is a frequency table holding the read counts, typically object[‘tab’]
q is the diversity order.
If dis =True, dissimilarities constrained between 0 and 1 are calculated, if dis =False, beta diversity constrained between 1 and 2 are calculated.
viewpoint can ‘local’ or ‘regional’, the difference is described in Chao et al. (2014), Annual Review of Ecology, Evolution, and Systematics, 45, 297-324.
diversity.phyl_beta(tab, tree, q=1, dis=True, viewpoint='local'):
Returns a dataframe of pairwise dissimilarities, which take a phylogenetic tree into account. The values are the phylogenetic beta diversity measures described by Chiu et al. (2014), Ecological Monographs, 84(1), 21-44.
tab is a frequency table holding the read counts, typically obj[‘tab’]
tree is a dataframe with phylogenetic tree information, typically obj[‘tree’].
q is the diversity order.
If dis =True, dissimilarities constrained between 0 and 1 are calculated, if dis =False, beta diversity constrained between 1 and 2 are calculated.
viewpoint can ‘local’ or ‘regional’.
diversity.func_beta(tab, distmat, q=1, dis=True, viewpoint='local'):
Returns a dataframe of pairwise dissimilarities, which take pairwise sequence distances into account. The values are the functional beta diversity measures described by Chiu et al. (2014), PLOS ONE, 9(7), e100014.
tab is a frequency table holding the read counts, typically obj[‘tab’].
distmat is a dataframe with pairwise distances between the sequences of OTUs/ASVs. The file can be generated with the stats.sequence_comparison() function.
q is the diversity order.
If dis =True, dissimilarities constrained between 0 and 1 are calculated, if dis =False, beta diversity constrained between 1 and 4 are calculated.
viewpoint can ‘local’ or ‘regional’.
diversity.jaccard(tab)
Returns a dataframe of pairwise Jaccard dissimilarities between samples.
tab is a count table holding the read counts, typically object[‘tab’]
Note that Jaccard dissimilarity is the same as naive_beta(tab, q=0, dis=True, viewpoint=’regional’)
diversity.bray(tab)
Returns a dataframe of pairwise Bray-Curtis dissimilarities between samples.
tab is a count table holding the read counts, typically obj[‘tab’]
diversity.naive_multi_beta(obj, var='None', q=1)
Returns a pandas dataframe containing taxonomic (naive) beta diversity, and local- and regional dissimilarity values for categories of samples.
obj is the qdiv object.
var is the column heading in the meta data used to categorize the samples. If a category has two or more samples, beta diversity for that category is calculated.
q is the diversity order.
diversity.phyl_multi_beta(obj, var='None', q=1)
Returns a pandas dataframe containing phylogenetic beta diversity, and local- and regional dissimilarity values for categories of samples.
obj is the qdiv object.
var is the column heading in the meta data used to categorize the samples. If a category has two or more samples, beta diversity for that category is calculated.
q is the diversity order.
diversity.func_multi_beta(obj, distmat, var='None', q=1)
Returns a pandas dataframe containing phylogenetic beta diversity, and local- and regional dissimilarity values for categories of samples.
obj is the qdiv object.
distmat is a distance matrix with pairwise distances between OTUs/ASVs, typically generated by the stats.sequence_comparison() function.
var is the column heading in the meta data used to categorize the samples. If a category has two or more samples, beta diversity for that category is calculated.
q is the diversity order.
diversity.dissimilarity_contributions(obj, var='None', q=1, divType='naive', index='local')
Returns a pandas dataframe with information about dissimilarity, number of samples, and the percentage contribution of each OTU/ASV to the observed dissimilarity.
obj is the qdiv object.
var is the column heading in the meta data used to categorize the samples. If a category has two or more samples, dissimilarity for that category is calculated.
q is the diversity order.
divType is the type of diversity: ‘naive’ or ‘phyl’.
index is the type of dissimilarity index (either local or regional)
Evenness¶
diversity.evenness(tab, tree='None', distmat='None', q=1, divType='naive', index='local', perspective='samples')
Returns evenness calculated according to Chao and Ricotta (2019) Ecology, 100(12), e02852. If q is 1, divType is ‘naive’, and index is either ‘CR1’, ‘CR2’, ‘local’ or ‘regional’, the evenness is identical to Pielou’s evenness.
tab is a count table, typically obj[‘tab’]
tree would typically be obj[‘tree’]. It must be specified only if divType=’phyl’.
distmat is a distance matrix. It must be specified only if divType=’func’.
q is diversity order.
divType is ‘naive’, ‘phyl’, or ‘func’.
index can be ‘CR1’, ‘CR2’, ‘CR3’, ‘CR4’, or ‘CR5’. These refer to the indices described in Table 1 in Chao and Ricotta (2019). index can also be ‘local’, which is equal to CR2, or ‘regional’, which is equal to CR1.
perspective can be ‘samples’, which means an evenness value is calculated for each sample (column) in the count table. perspective can also be ‘taxa’, which means an evenness value is calculated for each OTU/ASV (row) in the count table.