Merge branch 'latest' into release/v4.8.6

.git-blame-ignore-revs

@@ -0,0 +1,2 @@
+# Pre-commit updates (#2427)
+fee62922d8857ce93f1d4e90fd7240629d606997

.github/workflows/dev_envs.yml

@@ -5,25 +5,23 @@ on:
     branches: [latest]
 jobs:
   nix:
-    runs-on: ubuntu-latest
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ubuntu-latest, macos-14]
     steps:
     - uses: actions/checkout@v4
       with:
         fetch-depth: 0
 
-    - uses: cachix/install-nix-action@v25
-      with:
-        extra_nix_config: |
-          access-tokens = github.com=${{ secrets.GITHUB_TOKEN }}
-
-    - uses: cachix/cachix-action@v14
-      with:
-        name: sourmash-bio
-        authToken: '${{ secrets.CACHIX_AUTH_TOKEN }}'
+    - name: Install Nix
+      uses: DeterminateSystems/nix-installer-action@v4
+    - name: Run the Magic Nix Cache
+      uses: DeterminateSystems/magic-nix-cache-action@v1
 
     - run: nix run .# -- --version
 
-    - run: nix-shell --command "tox -e py310"
+    - run: nix develop --command bash -c "tox -e py310"
 
   mamba:
     runs-on: ubuntu-latest

doc/command-line.md

@@ -551,8 +551,8 @@ The sourmash `tax` or `taxonomy` commands integrate taxonomic
  `gather` command (we cannot combine separate `gather` runs for the
  same query). For supported databases (e.g. GTDB, NCBI), we provide
  taxonomy csv files, but they can also be generated for user-generated
- databases. As of v4.8, some sourmash taxonomy commands can also use `LIN`
- lineage information. For more information, see [databases](databases.md).
+ databases. As of v4.8 and 4.8.6, respectively, some sourmash taxonomy
+ commands can also use `LIN` or `ICTV` lineage information.
 
 `tax` commands rely upon the fact that `gather` provides both the total
  fraction of the query matched to each database matched, as well as a

flake.nix

@@ -30,40 +30,55 @@
           rustc = rustVersion;
         };
 
+        inherit (pkgs) lib;
+
         python = pkgs.python311Packages;
 
+        stdenv = if pkgs.stdenv.isDarwin then pkgs.overrideSDK pkgs.stdenv "11.0" else pkgs.stdenv;
+
+        commonArgs = {
+          src = ./.;
+          stdenv = stdenv;
+          preConfigure = lib.optionalString stdenv.isDarwin ''
+            export MACOSX_DEPLOYMENT_TARGET=10.14
+          '';
+
+          # Work around https://github.com/NixOS/nixpkgs/issues/166205.
+          env = lib.optionalAttrs stdenv.cc.isClang {
+            NIX_LDFLAGS = "-l${stdenv.cc.libcxx.cxxabi.libName}";
+          };
+
+          buildInputs = lib.optionals stdenv.isDarwin [ pkgs.libiconv pkgs.darwin.apple_sdk.frameworks.Security ];
+
+          nativeBuildInputs = with rustPlatform; [ cargoSetupHook maturinBuildHook bindgenHook ];
+        };
+
       in
 
       with pkgs;
       {
         packages = {
 
-          lib = rustPlatform.buildRustPackage {
+          lib = rustPlatform.buildRustPackage ( commonArgs // {
             name = "libsourmash";
-            src = lib.cleanSource ./.;
             copyLibs = true;
             cargoLock.lockFile = ./Cargo.lock;
             nativeBuildInputs = with rustPlatform; [ bindgenHook ];
-          };
+          });
 
-          sourmash = python.buildPythonPackage rec {
+          sourmash = python.buildPythonPackage ( commonArgs // rec {
             pname = "sourmash";
             version = "4.8.6";
             format = "pyproject";
 
-            src = ./.;
-
             cargoDeps = rustPlatform.importCargoLock {
               lockFile = ./Cargo.lock;
             };
 
-            nativeBuildInputs = with rustPlatform; [ cargoSetupHook maturinBuildHook bindgenHook ];
-
-            buildInputs = lib.optionals stdenv.isDarwin [ libiconv ];
             propagatedBuildInputs = with python; [ cffi deprecation cachetools bitstring numpy scipy matplotlib screed ];
 
             DYLD_LIBRARY_PATH = "${self.packages.${system}.lib}/lib";
-          };
+          });
 
           docker =
             let
@@ -83,7 +98,7 @@
 
         defaultPackage = self.packages.${system}.sourmash;
 
-        devShell = mkShell {
+        devShells.default = pkgs.mkShell.override { stdenv = stdenv; } (commonArgs // {
           nativeBuildInputs = with rustPlatform; [ bindgenHook ];
 
           buildInputs = [
@@ -113,15 +128,19 @@
             cargo-outdated
             cargo-udeps
             cargo-deny
-            cargo-semver-checks
+            #cargo-semver-checks
             nixpkgs-fmt
           ];
 
+          shellHook = ''
+              export MACOSX_DEPLOYMENT_TARGET=10.14
+            '';
+
           # Needed for matplotlib
           LD_LIBRARY_PATH = lib.makeLibraryPath [ pkgs.stdenv.cc.cc.lib ];
 
           # workaround for https://github.com/NixOS/nixpkgs/blob/48dfc9fa97d762bce28cc8372a2dd3805d14c633/doc/languages-frameworks/python.section.md#python-setuppy-bdist_wheel-cannot-create-whl
           SOURCE_DATE_EPOCH = 315532800; # 1980
-        };
+        });
       });
 }

src/core/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "sourmash"
-version = "0.12.0"
+version = "0.12.1"
 authors = ["Luiz Irber <luiz.irber@gmail.com>"]
 description = "MinHash sketches for genomic data"
 repository = "https://github.com/sourmash-bio/sourmash"

src/sourmash/cli/tax/annotate.py

@@ -78,6 +78,13 @@ def subparser(subparsers):
         default=False,
         help="use LIN taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain LIN lineage information.",
     )
+    subparser.add_argument(
+        "--ictv",
+        "--ictv-taxonomy",
+        action="store_true",
+        default=False,
+        help="use ICTV taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain ICTV ranks.",
+    )
 
 
 def main(args):

src/sourmash/cli/tax/genome.py

@@ -124,6 +124,13 @@ def subparser(subparsers):
         default=None,
         help="CSV containing 'name', 'lin' columns, where 'lin' is the lingroup prefix. Will restrict classification to these groups.",
     )
+    subparser.add_argument(
+        "--ictv",
+        "--ictv-taxonomy",
+        action="store_true",
+        default=False,
+        help="use ICTV taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain ICTV ranks.",
+    )
     add_tax_threshold_arg(subparser, 0.1)
     add_rank_arg(subparser)
 

src/sourmash/cli/tax/metagenome.py

@@ -116,6 +116,13 @@ def subparser(subparsers):
         default=None,
         help="CSV containing 'name', 'lin' columns, where 'lin' is the lingroup prefix. Will produce a 'lingroup' report containing taxonomic summarization for each group.",
     )
+    subparser.add_argument(
+        "--ictv",
+        "--ictv-taxonomy",
+        action="store_true",
+        default=False,
+        help="use ICTV taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain ICTV ranks.",
+    )
     add_rank_arg(subparser)
 
 

src/sourmash/cli/tax/summarize.py

@@ -57,6 +57,13 @@ def subparser(subparsers):
         default=False,
         help="use LIN taxonomy in place of standard taxonomic ranks.",
     )
+    subparser.add_argument(
+        "--ictv",
+        "--ictv-taxonomy",
+        action="store_true",
+        default=False,
+        help="use ICTV taxonomy in place of standard taxonomic ranks.  Note that the taxonomy CSV must contain ICTV ranks.",
+    )
 
 
 def main(args):

src/sourmash/tax/__main__.py

@@ -18,6 +18,7 @@
     RankLineageInfo,
     LINLineageInfo,
     AnnotateTaxResult,
+    ICTVRankLineageInfo,
 )
 
 usage = """
@@ -82,6 +83,7 @@ def metagenome(args):
             keep_identifier_versions=args.keep_identifier_versions,
             force=args.force,
             lins=args.lins,
+            ictv=args.ictv,
         )
         available_ranks = tax_assign.available_ranks
     except ValueError as exc:
@@ -113,6 +115,7 @@ def metagenome(args):
             keep_full_identifiers=args.keep_full_identifiers,
             keep_identifier_versions=args.keep_identifier_versions,
             lins=args.lins,
+            ictv=args.ictv,
         )
     except ValueError as exc:
         error(f"ERROR: {str(exc)}")
@@ -258,6 +261,7 @@ def genome(args):
             keep_identifier_versions=args.keep_identifier_versions,
             force=args.force,
             lins=args.lins,
+            ictv=args.ictv,
         )
         available_ranks = tax_assign.available_ranks
 
@@ -297,6 +301,7 @@ def genome(args):
             keep_full_identifiers=args.keep_full_identifiers,
             keep_identifier_versions=args.keep_identifier_versions,
             lins=args.lins,
+            ictv=args.ictv,
         )
 
     except ValueError as exc:
@@ -402,6 +407,7 @@ def annotate(args):
             keep_identifier_versions=args.keep_identifier_versions,
             force=args.force,
             lins=args.lins,
+            ictv=args.ictv,
         )
 
     except ValueError as exc:
@@ -466,6 +472,7 @@ def annotate(args):
                             raw=row,
                             id_col=id_col,
                             lins=args.lins,
+                            ictv=args.ictv,
                             keep_full_identifiers=args.keep_full_identifiers,
                             keep_identifier_versions=args.keep_identifier_versions,
                         )
@@ -591,6 +598,7 @@ def summarize(args):
             keep_full_identifiers=args.keep_full_identifiers,
             keep_identifier_versions=args.keep_identifier_versions,
             lins=args.lins,
+            ictv=args.ictv,
         )
     except ValueError as exc:
         error("ERROR while loading taxonomies!")
@@ -637,6 +645,8 @@ def summarize(args):
                 rank = lineage[-1].rank
                 if args.lins:
                     inf = LINLineageInfo(lineage=lineage)
+                elif args.ictv:
+                    inf = ICTVRankLineageInfo(lineage=lineage)
                 else:
                     inf = RankLineageInfo(lineage=lineage)
                 lin = inf.display_lineage()