make sure yeast gem works nicely

Project.toml

@@ -1,7 +1,7 @@
 name = "MATFBCModels"
 uuid = "bd2e13bf-42c0-49e1-9a9c-885d36daf88b"
 authors = ["The authors of MATFBCModels.jl"]
-version = "0.2.0"
+version = "1.0.0"
 
 [deps]
 AbstractFBCModels = "5a4f3dfa-1789-40f8-8221-69268c29937c"

src/constants.jl

@@ -33,6 +33,10 @@ metabolite_formulas = ["metFormula", "metFormulas"]
 
 metabolite_charges = ["metCharge", "metCharges"]
 
+metabolite_compartments = ["metCompartment", "metCompartments", "metComp", "metComps"]
+
+metabolite_compartment_tables = ["comps", "compNames"]
+
 reaction_names = ["rxnNames"]
 
 metabolite_names = ["metNames"]

src/interface.jl

@@ -108,7 +108,17 @@ function A.metabolite_charge(m::MATFBCModel, mid::String)
     parse_charge(m.mat[guesskeys(:metabolite_charges, m)][idx])
 end
 
-A.metabolite_compartment(m::MATFBCModel, mid::String) = nothing
+function A.metabolite_compartment(m::MATFBCModel, mid::String)::Union{Nothing,String}
+    comp_key = guesskeys_maybe(:metabolite_compartments, m)
+    isnothing(comp_key) && return nothing
+    res = m.mat[comp_key][findfirst(==(mid), A.metabolites(m))]
+    # now, if the metabolite is an integer or a (very integerish) float, it is an
+    # index to a table of metabolite names (such as in the yeast GEM)
+    typeof(res) <: Real || return parse_compartment(res)
+    ct_key = guesskeys_maybe(:metabolite_compartment_tables, m)
+    isnothing(ct_key) && return nothing
+    return parse_compartment(m.mat[ct_key][Int(res)])
+end
 
 function A.reaction_stoichiometry(m::MATFBCModel, rid::String)
     ridx = first(indexin([rid], m.mat[guesskeys(:reactions, m)]))[1] # get the index out of the cartesian index
@@ -178,7 +188,12 @@ function Base.convert(::Type{MATFBCModel}, m::A.AbstractFBCModel)
             ],
             "metFormulas" =>
                 [unparse_formula(A.metabolite_formula(m, mid)) for mid in A.metabolites(m)],
-            "metCharges" => [A.metabolite_charge(m, mid) for mid in A.metabolites(m)],
+            "metCharges" =>
+                [unparse_charge(A.metabolite_charge(m, mid)) for mid in A.metabolites(m)],
+            "metCompartments" => [
+                unparse_compartment(A.metabolite_compartment(m, mid)) for
+                mid in A.metabolites(m)
+            ],
         ),
     )
 end

src/utils.jl

@@ -1,5 +1,18 @@
 
-guesskeys(id, model) = first(intersect(keys(model.mat), getfield(key_names, id)))
+guesskeys_maybe(id, model) =
+    let keys = [k for k in getfield(key_names, id) if k in keys(model.mat)]
+        isempty(keys) ? nothing : first(keys)
+    end
+
+guesskeys(id, model) =
+    let key = guesskeys_maybe(id, model)
+        isnothing(key) ? throw(KeyError, "no applicable model key found for $id") : key
+    end
+
+function parse_compartment(x::String)
+    isempty(x) && return nothing
+    return x
+end
 
 function parse_formula(x::Maybe{String})
     isnothing(x) && return nothing
@@ -15,20 +28,26 @@ end
 
 function parse_charge(x)::Maybe{Int}
     if isa(x, Int)
-        x
+        return x
     elseif isa(x, Float64)
-        Int(x)
+        return isnan(x) ? nothing : Int(x)
     elseif isa(x, String)
-        Int(parse(Float64, x))
+        return Int(parse(Float64, x))
     elseif isnothing(x)
-        nothing
+        return nothing
     else
         throw(DomainError(x, "cannot parse charge"))
     end
 end
 
+unparse_compartment(::Nothing) = ""
+unparse_compartment(x::String) = x
+
 function unparse_formula(x::Maybe{A.MetaboliteFormula})
     isnothing(x) && return nothing
     ks = sort(collect(keys(x)))
     join(k * string(x[k]) for k in ks)
 end
+
+unparse_charge(::Nothing) = NaN
+unparse_charge(x::Int) = float(x)

test/runtests.jl

@@ -29,6 +29,12 @@ import SparseArrays
             "223db0b1ed69a7cc782f2a3093c1f48911a3c8bbd5bdf4bcdb13185cab5fdaa0",
             true,
         ),
+        (
+            "yeast-GEM",
+            "https://github.com/SysBioChalmers/yeast-GEM/raw/v8.6.2/model/yeast-GEM.mat",
+            "c2587e258501737e0141cd47e0f854a60a47faee2d4c6ad582a00e437676b181",
+            true,
+        ),
     ]
         path = joinpath(modeldir, "$name.mat")
         A.download_data_file(url, path, hash)