Support id list in download_alphafolds (#35)

This does less checking than the MSA-focused variant, but integrates
better with pipelines that incorporate other bioinformatic tools.

Author: timholy

src/alphafold.jl

@@ -76,7 +76,7 @@ end
 Query the [AlphaFold](https://alphafold.com/) API for the latest structure of `uniprotXname`.
 """
 function query_alphafold_latest(uniprotXname)
-    resp = HTTP.get("https://alphafold.com/api/prediction/$uniprotXname")
+    resp = HTTP.get("https://alphafold.com/api/prediction/$uniprotXname"; status_exception = false)
     if resp.status == 200
         j = JSON3.read(String(resp.body))[1]
         return j["pdbUrl"]
@@ -111,9 +111,13 @@ end
 
 """
     download_alphafolds(msa::AbstractMultipleSequenceAlignment; dirname=pwd())
+    download_alphafolds(ids; dirname=pwd())
 
 Download all available [AlphaFold](https://alphafold.com/) structures for the sequences in `msa`.
 Missing entries are silently skipped.
+
+If a `AbstractMultipleSequenceAlignment` is provided, the downloaded PDB file is checked to ensure that
+the residues in the MSA sequence match those in the PDB file. If they do not match, the PDB file is removed.
 """
 function download_alphafolds(msa::AbstractMultipleSequenceAlignment; dirname=pwd(), maxversion=nothing, kwargs...)
     maxversion === nothing || @warn "`download_alphafolds`: `maxversion` kwarg has no effect and is deprecated" maxlog=1
@@ -133,6 +137,18 @@ function download_alphafolds(msa::AbstractMultipleSequenceAlignment; dirname=pwd
     end
 end
 
+function download_alphafolds(ids; dirname=pwd(), kwargs...)
+    @showprogress 1 "Downloading AlphaFold files..." for uname in ids
+        url = query_alphafold_latest(uname)
+        url === nothing && continue
+        fn = split(url, '/')[end]
+        path = joinpath(dirname, fn)
+        if !isfile(path)
+            Downloads.download(url, path)
+        end
+    end
+end
+
 const pLDDTcolors = [  # see https://github.com/sokrypton/ColabFold?tab=readme-ov-file#faq
     RGB{N0f8}(0.051, 0.341, 0.827),
     RGB{N0f8}(0.416, 0.796, 0.945),

test/runtests.jl

@@ -372,6 +372,9 @@ using Test
                 conserved_cols = findall(msaentropies .< 0.5)
 
                 download_alphafolds(msa; dirname=path)
+                nfiles = length(readdir(path))
+                download_alphafolds(["K7N701", "G3UY47"]; dirname=path)
+                @test length(readdir(path)) == nfiles + 1
                 msacode2structfile = alphafoldfiles(msa, path)
                 afnbyidx(i) = getchain(joinpath(path, msacode2structfile[MSACode(sequencenames(msa)[i])]))