kallisto: attempt at making determinstic

2 files changed, 28 insertions, 0 deletions

diff --git a/tools/kallisto-quant-empty.h5 b/tools/kallisto-quant-empty.h5
new file mode 100644
index 0000000..2921292
--- /dev/null
+++ b/tools/kallisto-quant-empty.h5
diff --git a/tools/kallisto-quant.nix b/tools/kallisto-quant.nix
index 9f0ab57..c1083c8 100644
--- a/tools/kallisto-quant.nix
+++ b/tools/kallisto-quant.nix
@@ -22,6 +22,27 @@ inputs:
 let
   inherit (bionix.types) matchFiletype';
   isFastQ = matchFiletype' "kallisto-quant" {fq = _: true; gz = isFastQ; };
+
+  empty = ./kallisto-quant-empty.h5;
+
+  python = pkgs.python3Packages.python.withPackages (p: with p; [ h5py ]);
+
+  noStamp = pkgs.writeScript "nostamp.py" ''
+    #!${python}/bin/python
+    import h5py
+    import os
+    def copy(obj, out, path):
+      if type(obj) in [h5py._hl.group.Group,h5py._hl.files.File]:
+        for key in obj.keys():
+          if key != "start_time":
+            copy(obj[key], out, path + "/" + key)
+      elif type(obj)==h5py._hl.dataset.Dataset:
+        out.create_dataset(path, data=f[path], track_order=False, track_times = False)
+    with h5py.File(os.environ['out'] + "/abundance.h5", "r") as f:
+      with h5py.File("repack.h5", "a", track_order = False) as g:
+        copy(f, g, "")
+  '';
+
 in
 
 assert (all (x: isFastQ (x.filetype)) inputs);
@@ -43,6 +64,13 @@ stage {
       ${optionalString rfStranded "--rf-stranded"} \
       -t $NIX_BUILD_CORES \
       ${concatStringsSep " " inputs}
+
+    # Make deterministic by removing timestamps and using hdf5 empty template
+    cp ${empty} repack.h5
+    chmod 644 repack.h5
+    ${noStamp}
+    cp repack.h5 $out/abundance.h5
+    sed -i $out/run_info.json -e '/start_time/d'
   '';
   passthru.multicore = true;
 }