Humann profiles (#59)

* add demo data from humann

* add humann_profile function

* add humann_profiles function

* add tests, fix some stuff

Project.toml

@@ -13,6 +13,7 @@ Microbiome = "3bd8f0ae-a0f2-5238-a5af-e1b399a4940c"
 RCall = "6f49c342-dc21-5d91-9882-a32aef131414"
 SparseArrays = "2f01184e-e22b-5df5-ae63-d93ebab69eaf"
 Statistics = "10745b16-79ce-11e8-11f9-7d13ad32a3b2"
+Tables = "bd369af6-aec1-5ad0-b16a-f7cc5008161c"
 
 [compat]
 CSV = "0.8"

src/BiobakeryUtils.jl

@@ -6,6 +6,8 @@ export
     clean_abundance_tables,
     metaphlan_profile,
     metaphlan_profiles,
+    humann_profile,
+    humann_profiles,
     taxfilter,
     taxfilter!,
     parsetaxa,
@@ -29,4 +31,8 @@ include("general.jl")
 include("metaphlan.jl")
 include("humann.jl")
 
+function __init__()
+
+end
+
 end

src/humann.jl

@@ -1,3 +1,64 @@
+"""
+Stratified import currently non-functional
+"""
+function humann_profile(path::AbstractString; sample=basename(first(splitext(path))), stratified=false)
+    gfs = GeneFunction[]
+    abundances = Float64[]
+
+    for (i, (gf, abundance)) in enumerate(CSV.File(path, datarow=2, header=["function", "abundance"]))
+        if occursin('|', gf) # indicates a taxon-stratified entry
+            stratified || continue
+            # (gf, tax) = split(gf, '|')
+            # if tax == "unclassified"
+            #     tax = Taxon("unclassified")
+            # else
+            #     tm = match(r"s__(\w+)", tax)
+            #     cld = :species
+            #     if isnothing(tm)
+            #         tm = match(r"g__(\d+)", tax)
+            #         cld = :genus
+            #         isnothing(tm) && error("Incorrectly formatted taxon stratification: $tax")
+            #     end
+            #     tax = Taxon(string(tm.captures[1]), cld)
+            # end
+            # push!(gfs, GeneFunction(gf, tax))
+        else
+            push!(gfs, GeneFunction(gf))
+        end
+        push!(abundances, abundance)
+    end
+    mat = sparse(reshape(abundances, length(abundances), 1))
+    sample = sample isa Microbiome.AbstractSample ? sample : MicrobiomeSample(sample)
+
+    return CommunityProfile(mat, gfs, [sample])
+end
+
+"""
+Stratified import currently non-functional
+"""
+function humann_profiles(path::AbstractString; samples=nothing, stratified=false)
+    tbl = CSV.File(path)
+    gfs = GeneFunction[]
+    if !isnothing(samples) 
+        length(samples) == length(keys(first(tbl))) - 1 || throw(ArgumentError("Passed $(length(samples)) samples, but table has $(length(keys(first(tbl))) - 1)"))
+    else
+        samples = keys(first(tbl))[2:end]
+    end
+
+    # Need to add code to deal with stratified input
+    tbl = filter(row-> !occursin('|', row[1]), tbl)
+    mat = spzeros(length(tbl), length(samples))
+
+    for (i, (row)) in enumerate(tbl)
+        push!(gfs, GeneFunction(row[1]))
+        for j in 1:length(samples)
+            mat[i, j] = row[j+1]
+        end
+    end
+    samples = eltype(samples) == MicrobiomeSample ? samples : MicrobiomeSample.(string.(samples))
+    return CommunityProfile(mat, gfs, samples)
+end
+
 """
     function humann_regroup(df::AbstractDataFrame; inkind="uniref90", outkind::String="ec")