Add RDF HDT export stage

.editorconfig

@@ -0,0 +1,14 @@
+[*]
+end_of_line = lf
+charset = utf-8
+trim_trailing_whitespace = true
+insert_final_newline = true
+indent_style = space
+indent_size = 2
+max_line_length = 120
+
+[Makefile]
+indent_style = tab
+
+[*.{diff,md}]
+trim_trailing_whitespace = false

.envrc

@@ -0,0 +1 @@
+use flake

.gitignore

@@ -1,5 +1,12 @@
+.*
+!.editorconfig
+!.envrc
+!.github
+!.gitignore
+
 logs
 /download
 /temp
 /brick
-/raw
+/raw
+/rdf

dvc.lock

@@ -49,8 +49,8 @@ stages:
       nfiles: 337
     - path: stages/3_write_brick.R
       hash: md5
-      md5: c5dbdeded250cf30b3af70cc436cb77f
-      size: 942
+      md5: 1377f37bad12cec9d4a0e8afae48140d
+      size: 1159
     - path: temp
       hash: md5
       md5: 01a6bc69d09016a79132483c28820178.dir
@@ -59,13 +59,39 @@ stages:
     outs:
     - path: brick/tox21.parquet
       hash: md5
-      md5: 247348fd1254be594b21d202d514d55d
-      size: 1231260948
+      md5: 0eb67f9b2fc3b16aff313d8ea9cdc359
+      size: 1237116063
     - path: brick/tox21_aggregated.parquet
       hash: md5
       md5: 4de160eb524d30b610ed369d19776af8
       size: 34091575
     - path: brick/tox21lib.parquet
       hash: md5
-      md5: 0f67d0d71773dfff4e9658334b1c8895
+      md5: be3b8da4eebac921c7eca80177280e93
       size: 825475
+  rml:
+    cmd: bash stages/4_rml.sh
+    deps:
+    - path: brick/
+      hash: md5
+      md5: 268c1928221d9976b99ecaedc9edf48f.dir
+      size: 1272033113
+      nfiles: 3
+    - path: morph-kgc.ini
+      hash: md5
+      md5: 4db7b2ab3384fb468b716ddc7161a00d
+      size: 132
+    - path: stages/4_rml.sh
+      hash: md5
+      md5: 1a6f8e5e301c61b24fc3023ad970d2e8
+      size: 112
+    - path: tox21.rml.ttl
+      hash: md5
+      md5: 6279d0e422d2cb7a33ae813afe3611d3
+      size: 2121
+    outs:
+    - path: rdf/
+      hash: md5
+      md5: 7e612fd7f1b73c26ddf5a1a2caee495f.dir
+      size: 355139697
+      nfiles: 1

dvc.yaml

@@ -8,15 +8,15 @@ stages:
     outs:
       - download
       - temp
-  unzip: 
+  unzip:
     cmd: stages/2_unzip.sh
     deps:
       - stages/2_unzip.sh
       - download
       - temp
     outs:
       - raw
-  build: 
+  build:
     cmd: Rscript stages/3_write_brick.R
     deps:
       - stages/3_write_brick.R
@@ -26,3 +26,12 @@ stages:
       - brick/tox21.parquet
       - brick/tox21_aggregated.parquet
       - brick/tox21lib.parquet
+  rml:
+    cmd: bash stages/4_rml.sh
+    deps:
+      - brick/
+      - stages/4_rml.sh
+      - tox21.rml.ttl
+      - morph-kgc.ini
+    outs:
+      - rdf/

flake.nix

@@ -0,0 +1,35 @@
+{
+  description = "Tox21 biobrick";
+
+  inputs = {
+    nixpkgs.url = "https://flakehub.com/f/NixOS/nixpkgs/*.tar.gz";
+    flake-utils.url = "https://flakehub.com/f/numtide/flake-utils/*.tar.gz";
+    biobricks-R = {
+      url = "github:biobricks-ai/biobricks-R";
+      inputs.flake-utils.follows = "flake-utils";
+      inputs.nixpkgs.follows = "nixpkgs";
+    };
+    hdt-cpp = {
+      url = "github:insilica/nix-hdt";
+      inputs.flake-utils.follows = "flake-utils";
+      inputs.nixpkgs.follows = "nixpkgs";
+    };
+    morph-kgc = {
+      url = "github:insilica/nix-morph-kgc";
+      inputs.flake-utils.follows = "flake-utils";
+      inputs.nixpkgs.follows = "nixpkgs";
+    };
+  };
+
+  outputs = { self, nixpkgs, flake-utils, biobricks-R, hdt-cpp, morph-kgc }:
+    flake-utils.lib.eachDefaultSystem (system:
+      with import nixpkgs { inherit system; }; {
+        devShells.default = mkShell {
+          buildInputs = [
+            biobricks-R.packages.${system}.rEnv
+            hdt-cpp.packages.${system}.default
+            morph-kgc.packages.${system}.default
+          ];
+        };
+      });
+}