Distance metrics (#57)

* Add jaccard and hellinger distances.

* Update documentation file with new distances

* Increase version number to 0.8.0

* Bump version number in docs/Project.toml

* Change version to 0.7.1

Project.toml

@@ -4,7 +4,7 @@ keywords = ["microbiology", "microbiome", "biology"]
 license = "MIT"
 desc = "Functions and types for working with microbial community data"
 authors = ["@kescobo <kevbonham@gmail.com>"]
-version = "0.7.0"
+version = "0.7.1"
 
 [deps]
 AxisIndices = "f52c9ee2-1b1c-4fd8-8546-6350938c7f11"

docs/Project.toml

@@ -4,4 +4,4 @@ Microbiome = "3bd8f0ae-a0f2-5238-a5af-e1b399a4940c"
 
 [compat]
 Documenter = "0.24"
-Microbiome = "0.7"
+Microbiome = "0.7.1"

docs/src/diversity.md

@@ -16,7 +16,7 @@ Quite often, it's useful to boil stuff down to distances between samples.
 from [`Distances.jl`](https://github.com/JuliaStats/Distances.jl)
 to get a symetric distance matrix.
 
-Right now, only Bray-Curtis is implemented, because it's the only one I use,
+Right now, only Bray-Curtis, Jaccard, and Hellinger are implemented, 
 but it would be straightforward to add any others.
 Open an issue if you want them!
 
@@ -25,5 +25,7 @@ using the `fit(MDS, ...)` from `MultivariateStats.jl` under the hood.
 
 ```@docs
 braycurtis
+jaccard
+hellinger
 pcoa
 ```

src/Microbiome.jl

@@ -53,6 +53,8 @@ export ginisimpson,
        present,
        prevalence,
        braycurtis,
+       jaccard,
+       hellinger,
        pcoa
 
 using Statistics

src/diversity.jl

@@ -79,15 +79,28 @@ end
 """
     braycurtis(abt::AbstractAbundanceTable)
 
-Retruns a pairwise Bray-Curtis dissimilarity matrix. 
+Returns a pairwise Bray-Curtis dissimilarity matrix.
 """
 braycurtis(abt::AbstractAbundanceTable) = pairwise(BrayCurtis(), collect(abundances(abt)), dims=2)
 
+"""
+    jaccard(abt::AbstractAbundanceTable)
+
+Returns a pairwise Jaccard distance matrix.
+"""
+jaccard(abt::AbstractAbundanceTable) = pairwise(Jaccard(), collect(abundances(abt)), dims=2)
+
+"""
+    hellinger(abt::AbstractAbundanceTable)
+
+Returns a pairwise Hellinger distance matrix.
+"""
+hellinger(abt::AbstractAbundanceTable) = pairwise(HellingerDist(), collect(abundances(abt)), dims=2)
 
 """
     pcoa(abt::AbstractAbundanceTable, f=braycurtis) 
 
-Returns eigenvectors from fiting `MDS` to a distance metric generated by `f`,
+Returns eigenvectors from fitting `MDS` to a distance metric generated by `f`,
 by default `braycurtis`.
 """
-pcoa(abt::AbstractAbundanceTable, f=braycurtis) = fit(MDS, f(abt), distances = true)
+pcoa(abt::AbstractAbundanceTable, f=braycurtis) = fit(MDS, f(abt), distances = true)

test/runtests.jl

@@ -192,6 +192,8 @@ end
         end
 
         @test all(braycurtis(comm) .== [0 1 1 1 1; 1 0 1 1 1; 1 1 0 1 1; 1 1 1 0 1; 1 1 1 1 0])
+        @test all(jaccard(comm) .== [0 1 1 1 1; 1 0 1 1 1; 1 1 0 1 1; 1 1 1 0 1; 1 1 1 1 0])
+        @test all(hellinger(comm) .== [0 1 1 1 1; 1 0 1 1 1; 1 1 0 1 1; 1 1 1 0 1; 1 1 1 1 0])
         @test pcoa(comm) isa MDS
     end