Additional tests on distance matrix (#56)

* 🐛fix the taxodistance function (last taxa self-distance was 0)

* ✅ taxonomicdistance

* ⚡ taxonomicdistance benchmark

* ✅ authority

* 🩹 v0.4.1

Project.toml

@@ -1,7 +1,7 @@
 name = "NCBITaxonomy"
 uuid = "f88b31d2-eb98-4433-b52d-2dd32bc6efce"
 authors = ["Timothée Poisot <[email protected]>"]
-version = "0.4.0"
+version = "0.4.1"
 
 [deps]
 AbstractTrees = "1520ce14-60c1-5f80-bbc7-55ef81b5835c"

benchmark/benchmarks.jl

@@ -2,6 +2,17 @@ using BenchmarkTools
 using NCBITaxonomy
 using AbstractTrees
 
+# Setup
+tax = [
+    ncbi"Lamellodiscus",
+    ncbi"Dactylogyrus",
+    ncbi"Paradiplozoon",
+    ncbi"Diplectanum",
+    ncbi"Echinoplectanum",
+]
+
+mf = mammalfilter()
+
 const SUITE = BenchmarkGroup()
 
 # Construction of name finders
@@ -14,9 +25,6 @@ SUITE["name finders"]["mammals (inclusive)"] = @benchmarkable mammalfilter(true)
 
 SUITE["name finders"]["phage"] = @benchmarkable phagefilter()
 
-# Prepare a mammal filter next
-mf = mammalfilter()
-
 # Ability to locate taxa
 
 SUITE["taxon search"] = BenchmarkGroup(["namefinding", "search"])
@@ -49,10 +57,16 @@ SUITE["traversal"]["lineage"] = @benchmarkable lineage(ncbi"Lamellodiscus ignora
 SUITE["traversal"]["common ancestor (pair)"] =
     @benchmarkable commonancestor(ncbi"Lamellodiscus", ncbi"Dactylogyrus")
 
-SUITE["traversal"]["common ancestor (array)"] = @benchmarkable commonancestor([
-    ncbi"Lamellodiscus",
-    ncbi"Dactylogyrus",
-    ncbi"Paradiplozoon",
-    ncbi"Diplectanum",
-    ncbi"Echinoplectanum",
-])
+SUITE["traversal"]["common ancestor (array)"] = @benchmarkable commonancestor(tax)
+
+# Utility functions
+
+SUITE["utilities"] = BenchmarkGroup(["utilities", "quality of life"])
+
+SUITE["utilities"]["distance matrix (no allocation)"] =
+    @benchmarkable taxonomicdistance(tax)
+
+D = zeros(Float64, (length(tax), length(tax)))
+
+SUITE["utilities"]["distance matrix (pre-allocation)"] =
+    @benchmarkable taxonomicdistance!(D, tax)

src/utility/taxonomicdistance.jl

@@ -54,6 +54,8 @@ function taxonomicdistance!(
             D[i, j] = D[j, i] = minimum([get(d, s, d[:fallback]) for s in shared])
         end
     end
+    l = length(tax)
+    D[l, l] = strict ? get(d, rank(tax[l]), d[:fallback]) : 0.0
     return D
 end
 

test/synonyms.jl

@@ -1,9 +1,12 @@
 module TestSynonymsIssues
 
-    using Test
-    using NCBITaxonomy
+using Test
+using NCBITaxonomy
 
-    @test_throws NameHasMultipleMatches taxon("gorilla"; casesensitive=false)
-    @test isa(taxon("gorilla"; casesensitive=false, preferscientific=true), NCBITaxon)
+@test_throws NameHasMultipleMatches taxon("gorilla"; casesensitive = false)
+@test isa(taxon("gorilla"; casesensitive = false, preferscientific = true), NCBITaxon)
+
+@test authority(ncbi"Lamellodiscus kechemirae") ==
+      "Lamellodiscus kechemirae Amine & Euzet, 2005"
 
 end

test/taxodistance.jl

@@ -0,0 +1,19 @@
+module TestTaxoDistance
+using NCBITaxonomy
+using Test
+
+tax = [
+    ncbi"Lamellodiscus",
+    ncbi"Dactylogyrus",
+    ncbi"Paradiplozoon",
+    ncbi"Diplectanum",
+    ncbi"Echinoplectanum",
+]
+
+TD = taxonomicdistance(tax)
+
+ETD = Float64[1 3 4 2 2; 3 1 4 3 3; 4 4 1 4 4; 2 3 4 1 2; 2 3 4 2 1]
+
+@test TD == ETD
+
+end