From ce10b2db7c067b16cc3a683073a175fbf901ff4e Mon Sep 17 00:00:00 2001
From: anton083 <anton.oresten42@gmail.com>
Date: Tue, 15 Oct 2024 17:15:11 +0200
Subject: [PATCH 1/2] Add mmcifutils.jl

---
 src/io/io.jl         |  1 +
 src/io/mmcifutils.jl | 47 ++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 48 insertions(+)
 create mode 100644 src/io/mmcifutils.jl

diff --git a/src/io/io.jl b/src/io/io.jl
index 9e42e96..91092be 100644
--- a/src/io/io.jl
+++ b/src/io/io.jl
@@ -18,3 +18,4 @@ include("renumber.jl")
 include("read.jl")
 include("write.jl")
 include("download.jl")
+include("mmcifutils.jl")
diff --git a/src/io/mmcifutils.jl b/src/io/mmcifutils.jl
new file mode 100644
index 0000000..f95f8f6
--- /dev/null
+++ b/src/io/mmcifutils.jl
@@ -0,0 +1,47 @@
+export getmmcif
+export mapmmcif
+
+function map_first_occurrence(u, v)
+    d = Dict{eltype(u),eltype(v)}()
+    for (x, y) in zip(u, v)
+        haskey(d, x) || (d[x] = y)
+    end
+    d
+end
+
+map_last_occurrence(u, v) = Dict(zip(u, v))
+
+compose_map(d1, d2, fallback="?") = Dict(k => get(d2, v, fallback) for (k,v) in d1)
+
+getmmcif(mmcifdict::AbstractDict{String,Vector{String}}, key::AbstractString) = get(mmcifdict, key, String[])
+
+"""
+    mapmmcif(mmcifdict, field1 => field2, field3 => field4, ...)
+
+```jldoctest
+julia> import BioStructures
+
+julia> filename BioStructures.downloadpdb("3HFM", format=BioStructures.MMCIFFormat);
+[ Info: Downloading file from PDB: 3HFM
+
+julia> mmcifdict = BioStructures.MMCIFDict(filename);
+
+julia> mapmmcif(mmcifdict,
+           "_atom_site.auth_asym_id"   => "_atom_site.label_entity_id",
+           "_entity_src_gen.entity_id" => "_entity_src_gen.pdbx_gene_src_ncbi_taxonomy_id")
+Dict{String, String} with 3 entries:
+  "Y" => "9031"
+  "L" => "10090"
+  "H" => "10090"
+```
+"""
+mapmmcif(mmcifdict, pairs::Pair{String,String}...) =
+    mapreduce(((from,to),) -> map_first_occurrence(getmmcif(mmcifdict, from), getmmcif(mmcifdict, to)), compose_map, pairs)
+
+get_auth_asym_to_entity(mmcifdict) = mapmmcif(mmcifdict, "_atom_site.auth_asym_id" => "_atom_site.label_entity_id")
+
+function get_auth_asym_to_taxid(mmcifdict)
+    mapmmcif(mmcifdict,
+        "_atom_site.auth_asym_id"   => "_atom_site.label_entity_id",
+        "_entity_src_gen.entity_id" => "_entity_src_gen.pdbx_gene_src_ncbi_taxonomy_id")
+end

From c61fc99239430c620360bd8ab0ee4940b3f404f5 Mon Sep 17 00:00:00 2001
From: anton083 <anton.oresten42@gmail.com>
Date: Thu, 17 Oct 2024 17:02:24 +0200
Subject: [PATCH 2/2] Add test, export BioStructures

---
 src/io/io.jl     | 4 +++-
 test/runtests.jl | 8 ++++++++
 2 files changed, 11 insertions(+), 1 deletion(-)

diff --git a/src/io/io.jl b/src/io/io.jl
index 91092be..5f606a8 100644
--- a/src/io/io.jl
+++ b/src/io/io.jl
@@ -1,4 +1,6 @@
-using BioStructures: BioStructures, PDBFormat, MMCIFFormat
+using BioStructures: BioStructures, MMCIFDict, PDBFormat, MMCIFFormat
+
+export BioStructures, MMCIFDict, PDBFormat, MMCIFFormat
 
 const ProteinFileFormat = Union{PDBFormat, MMCIFFormat}
 const AMINOACIDS = Set("ACDEFGHIKLMNPQRSTVWY")
diff --git a/test/runtests.jl b/test/runtests.jl
index 3a92eab..4c00c4e 100644
--- a/test/runtests.jl
+++ b/test/runtests.jl
@@ -77,6 +77,14 @@ using Test
             @test chains[1].sequence == new_chains[1].sequence
         end
 
+        @testset "mmcifutils" begin
+            mktempdir() do dir
+                structure = pdbentry("3HFM"; dir)
+                mmcifdict = MMCIFDict(joinpath(dir, structure.name))
+                @test ProteinChains.get_auth_asym_to_taxid(mmcifdict) == Dict("Y" => "9031", "L" => "10090", "H" => "10090")
+            end
+        end
+
     end
 
     @testset "store" begin