diff --git a/DNAnexus/Makefile b/DNAnexus/Makefile
new file mode 100644
index 0000000..6f9e34c
--- /dev/null
+++ b/DNAnexus/Makefile
@@ -0,0 +1,15 @@
+all: default
+
+default:
+	rm -rf resources/PhyloCSF*
+	mkdir -p resources
+	cp -r ../PhyloCSF ../PhyloCSF.Linux.x86_64 ../PhyloCSF_Parameters resources
+	$(MAKE) -C src all
+
+clean:
+	$(MAKE) -C src clean
+
+test: all
+	test/test.sh
+
+.PHONY: all default clean test
diff --git a/DNAnexus/Readme.developer.md b/DNAnexus/Readme.developer.md
new file mode 100644
index 0000000..7a0f176
--- /dev/null
+++ b/DNAnexus/Readme.developer.md
@@ -0,0 +1,35 @@
+# PhyloCSF Developer Readme
+
+<!--
+TODO: Please edit this Readme.developer.md file to include information
+for developers or advanced users, for example:
+
+* Information about app internals and implementation details
+* How to report bugs or contribute to development
+-->
+
+## Running this app with additional computational resources
+
+This app has the following entry points:
+
+* main
+* process
+* postprocess
+
+When running this app, you can override the instance type to be used for each
+entry point by providing the ``systemRequirements`` field to
+```/applet-XXXX/run``` or ```/app-XXXX/run```, as follows:
+
+    {
+      systemRequirements: {
+        "main": {"instanceType": "dx_m1.large"},
+        "process": {"instanceType": "dx_m1.large"},
+        "postprocess": {"instanceType": "dx_m1.large"}
+      },
+      [...]
+    }
+
+See <a
+href="http://wiki.dnanexus.com/API-Specification-v1.0.0/IO-and-Run-Specifications#Run-Specification">Run
+Specification</a> in the API documentation for more information about the
+available instance types.
diff --git a/DNAnexus/Readme.md b/DNAnexus/Readme.md
new file mode 100644
index 0000000..8a4a00e
--- /dev/null
+++ b/DNAnexus/Readme.md
@@ -0,0 +1,51 @@
+<!-- dx-header -->
+# PhyloCSF (DNAnexus Platform App)
+
+Phylogenetic analysis of multi-species genome sequence alignments to identify conserved protein-coding regions
+
+This is the source code for an app that runs on the DNAnexus Platform.
+For more information about how to run or modify it, see
+http://wiki.dnanexus.com/.
+<!-- /dx-header -->
+
+
+
+<!--
+TODO: This app directory was automatically generated by dx-app-wizard;
+please edit this Readme.md file to include essential documentation about
+your app that would be helpful to users. (Also see the
+Readme.developer.md.) Once you're done, you can remove these TODO
+comments.
+
+For more info, see http://wiki.dnanexus.com/Developer-Portal.
+-->
+
+<!--
+TODO: Fill in additional info about how to use each input and output
+below.
+-->
+
+## Inputs
+
+* **alignments** the ``CrossSpeciesAlignments`` to evaluate, usually generated by the MAF Stitcher app
+* **output_name** name of the output table
+* **alignments_per_job** controls the degree of parallelization
+* **species_set** usually auto-detected for alignments produced by the MAF Stitcher app
+* **frames** search over three or six reading frames
+* **orf** search over ORFs within each alignment
+* **min_codons** minimum length considered in ORF search mode (with ``orf != AsIs``)
+
+## Advanced Inputs
+
+* **strategy** model and parameter optimization strategy
+* **all_scores** report all scores computed (with ``frames != 1`` or ``orf != AsIs``)
+* **bls** include a Branch Length Score along with PhyloCSF score in output table
+* **anc_comp** include an ancestral composition score along with PhyloCSF score in output table
+* **dna** include evaluated DNA sequence in output table
+* **aa** include evaluated amino acid sequence in output table
+* **instance_type** instance type for parallel jobs
+
+## Outputs
+
+* **scores**
+* **errors** number of alignments without scores (see job log for details)
diff --git a/DNAnexus/dxapp.json b/DNAnexus/dxapp.json
new file mode 100644
index 0000000..0f8fdd5
--- /dev/null
+++ b/DNAnexus/dxapp.json
@@ -0,0 +1,137 @@
+{
+  "name": "PhyloCSF",
+  "title": "PhyloCSF",
+  "summary": "Phylogenetic analysis of multi-species genome sequence alignments to identify conserved protein-coding regions",
+  "dxapi": "1.0.0",
+  "version": "0.1.0",
+  "inputSpec": [
+    {
+      "name": "alignments",
+      "class": "gtable",
+      "type": "CrossSpeciesAlignments",
+      "optional": false,
+      "help": "Alignments to evaluate, usually generated by the MAF Stitcher app"
+    },
+    {
+      "name": "output_name",
+      "class": "string",
+      "optional": true,
+      "help": "Name of output table"
+    },
+    {
+      "name": "alignments_per_job",
+      "class": "int",
+      "optional": true,
+      "default": 500,
+      "help": "Degree of parallelization"
+    },
+    {
+      "name": "species_set",
+      "class": "string",
+      "choices": ["12flies", "29mammals"],
+      "optional": true,
+      "help": "Usually auto-detected for alignments produced by the MAF Stitcher app"
+    },
+    {
+      "name": "frames",
+      "class": "int",
+      "choices": [1,3,6],
+      "optional": true,
+      "default": 1,
+      "help": "Search over three or six reading frames"
+    },
+    {
+      "name": "orf",
+      "class": "string",
+      "choices": ["AsIs", "ATGStop", "StopStop", "StopStop3", "ToFirstStop", "FromLastStop", "ToOrFromStop"],
+      "optional": true,
+      "default": "AsIs",
+      "help": "Search over ORFs within each alignment (usually with frames=3 or frames=6)"
+    },
+    {
+      "name": "min_codons",
+      "class": "int",
+      "optional": true,
+      "default": 25,
+      "help": "Minimum length considered in ORF search mode (with orf != AsIs)"
+    },
+    {
+      "name": "strategy",
+      "class": "string",
+      "choices": ["mle", "fixed", "omega"],
+      "optional": true,
+      "default": "mle",
+      "group": "Advanced",
+      "help": "Model and parameter optimization strategy"
+    },
+    {
+      "name": "all_scores",
+      "class": "boolean",
+      "optional": true,
+      "default": false,
+      "group": "Advanced",
+      "help": "Report all scores computed (with frames != 1 or orf != AsIs)"
+    },
+    {
+      "name": "bls",
+      "class": "boolean",
+      "optional": true,
+      "default": false,
+      "group": "Advanced",
+      "help": "Include a Branch Length Score along with PhyloCSF score in output table"
+    },
+    {
+      "name": "anc_comp",
+      "class": "boolean",
+      "optional": true,
+      "default": false,
+      "group": "Advanced",
+      "help": "Include an ancestral composition score along with PhyloCSF score in output table"
+    },
+    {
+      "name": "dna",
+      "class": "boolean",
+      "optional": true,
+      "default": false,
+      "group": "Advanced",
+      "help": "Include evaluted DNA sequence in output table"
+    },
+    {
+      "name": "aa",
+      "class": "boolean",
+      "optional": true,
+      "default": false,
+      "group": "Advanced",
+      "help": "Include evaluated amino acid sequence in output table"
+    },
+    {
+      "name": "instance_type",
+      "class": "string",
+      "choices": ["dx_m1.large", "dx_c1.xlarge"],
+      "optional": true,
+      "group": "Advanced",
+      "help": "Instance type for parallel jobs"
+    }
+  ],
+  "outputSpec": [
+    {
+      "name": "scores",
+      "class": "gtable"
+    },
+    {
+      "name": "errors",
+      "class": "int"
+    }
+  ],
+  "runSpec": {
+    "interpreter": "bash",
+    "file": "src/run.sh",
+    "execDepends": [ {"name": "gsl-bin"} ]
+  },
+  "details": {
+    "upstreamUrl": "https://github.com/mlin/PhyloCSF",
+    "citations": [
+      "doi:10.1093/bioinformatics/btr209"
+    ]
+  }
+}
diff --git a/DNAnexus/src/Makefile b/DNAnexus/src/Makefile
new file mode 100644
index 0000000..1e31d73
--- /dev/null
+++ b/DNAnexus/src/Makefile
@@ -0,0 +1,12 @@
+all:
+	mkdir -p ../resources
+	$(MAKE) ../resources/dxPhyloCSF
+
+../resources/dxPhyloCSF: _tags dxPhyloCSF.ml
+	ocamlbuild -use-ocamlfind dxPhyloCSF.native
+	mkdir -p ../resources
+	cp -f dxPhyloCSF.native ../resources/dxPhyloCSF
+
+clean:
+	ocamlbuild -clean
+	rm -f ../resources/dxPhyloCSF
diff --git a/DNAnexus/src/_tags b/DNAnexus/src/_tags
new file mode 100644
index 0000000..15164b1
--- /dev/null
+++ b/DNAnexus/src/_tags
@@ -0,0 +1,2 @@
+<*.ml{,i}>: pp(ocaml+twt)
+true: package(DNAnexus), package(forkwork), thread, debug
diff --git a/DNAnexus/src/dxPhyloCSF.ml b/DNAnexus/src/dxPhyloCSF.ml
new file mode 100644
index 0000000..1655721
--- /dev/null
+++ b/DNAnexus/src/dxPhyloCSF.ml
@@ -0,0 +1,316 @@
+open Batteries
+open Printf
+open JSON.Operators
+open DNAnexus
+
+Printexc.record_backtrace true
+
+(* abbreviated JSON casts *)
+module J = struct
+  include JSON
+  let of_str_list lst = of_list (List.map (fun x -> `String x) lst)
+  let of_str_assoc lst = of_assoc (List.map (fun (k,v) -> k, `String v) lst)
+
+(* return a list of (starting,limit) tuples *)
+let partition_rows length rows_per_job =  
+  if length = 0 then []
+  else
+    (* make distribution more uniform *)
+    let jobs = 1 + (length - 1) / rows_per_job
+    let k = int_of_float (ceil (float length /. float jobs))
+    (0 --^ jobs) /@ (fun i -> (i*k), (min k (length-i*k))) |> List.of_enum
+
+let make_scores_gtable name app_id input =
+  let optional_colspecs =
+    List.filter_map
+      fun (k,ty) ->
+        if J.bool (input$k) then Some (k,ty)
+        else None
+      ["bls",`Float; "anc_comp",`Float; "dna",`String; "aa",`String]
+  let frames = J.int (input$"frames")
+  let optional_colspecs =
+    if frames <> 6 then optional_colspecs
+    else ("strand",`String) :: optional_colspecs
+  let optional_colspecs =
+    if frames = 1 && J.string (input$"orf") = "AsIs" then optional_colspecs
+    else ("lo",`Int64) :: ("hi",`Int64) :: optional_colspecs
+  GTable.make_new (J.of_assoc [
+    "name", `String name;
+    "columns", GTable.json_of_columns (Array.of_list (("name",`String) :: ("score",`Float) :: optional_colspecs));
+    "indices", GTable.json_of_indices ["name", `Lexicographic ["name",`Asc,None]];
+    "details", J.of_assoc [
+      "PhyloCSF", make_link app_id;
+      "PhyloCSF_input", input
+    ]
+  ])
+
+let get_species_names alignments_table = 
+  try
+    let alignments_cols = GTable.columns alignments_table
+    let n = Array.length alignments_cols
+    assert (n >= 4)
+    Array.sub alignments_cols 2 (n-2) |> Array.map
+      function
+        | (sp,`String) when sp <> "" -> sp
+        | _ -> assert false
+  with 
+    | exn ->
+      eprintf "%s\n%s\n" (Printexc.to_string exn) (Printexc.get_backtrace ())
+      raise (DNAnexus.AppInternalError "Could not understand input alignments table. Please make sure it was generated with the maf_stitcher app.")
+
+type configuration = {
+  strategy : string;
+  frames : int;
+  orf : string;
+  min_codons : int;
+  all_scores : bool;
+  bls : bool;
+  anc_comp : bool;
+  dna : bool;
+  aa : bool
+}
+
+let run_PhyloCSF species_set species_names alignment_row cfg =
+  let n = Array.length species_names
+  assert (Array.length alignment_row = n+3)
+  let alignment_name = match alignment_row.(1) with
+    | `String nm -> nm
+    | _ -> assert false
+  let seqs =
+    Array.sub alignment_row 3 n |> Array.map
+      function
+        | `String sp -> sp
+        | _ -> assert false
+
+  (* Formulate PhyloCSF command line and start process *)
+  (* TODO: keep PhyloCSF online to avoid reinstantiating phylo-models for every alignment *)
+  (* TODO: support --species *)
+  let cmd =
+    sprintf "/PhyloCSF %s --removeRefGaps --strategy=%s --frames=%d --orf=%s --minCodons=%d%s%s%s%s%s"
+      species_set
+      cfg.strategy
+      cfg.frames
+      cfg.orf
+      cfg.min_codons
+      if cfg.all_scores then " --allScores" else ""
+      if cfg.bls then " --bls" else ""
+      if cfg.anc_comp then " --ancComp" else ""
+      if cfg.dna then " --dna" else ""
+      if cfg.aa then " --aa" else ""
+
+  let phylocsf_out, phylocsf_in = Unix.open_process cmd
+
+  (* write Multi-FASTA alginment into PhyloCSF *)
+  for i = 0 to n-1 do
+    fprintf phylocsf_in ">%s\n" species_names.(i)
+    fprintf phylocsf_in "%s\n" seqs.(i)
+  flush phylocsf_in
+  close_out phylocsf_in
+
+  (* collect output *)
+  let result = IO.read_all phylocsf_out
+
+  match Unix.close_process (phylocsf_out, phylocsf_in) with 
+    | Unix.WEXITED 0 -> ()
+    | _ ->
+        printf "%s\n" result
+        raise (DNAnexus.AppInternalError ("PhyloCSF exited abnormally; see the job log for more details."))
+
+  (* for each output line *)
+  List.of_enum
+    IO.lines_of (IO.input_string result) |> Enum.filter_map
+      fun line ->
+        (* ignore comments *)
+        if String.length line = 0 || line.[0] = '#' then
+          print_endline line
+          None
+        else
+          (* parse line, extract score *)
+          let flds = Array.of_list (String.nsplit line "\t")
+          assert (Array.length flds >= 2)
+          assert (flds.(1) = "score(decibans)" || flds.(1) = "orf_score(decibans)" || flds.(1) = "max_score(decibans)")
+          if cfg.all_scores && flds.(1) = "max_score(decibans)" then (* these are redundant *)
+            None
+          else
+            let sc = float_of_string flds.(2)
+            (* extract optional outputs *)
+            let i = ref 2
+            let extras = ref []
+            (* lo & hi *)
+            if cfg.frames <> 1 || cfg.orf <> "AsIs" then
+              incr i
+              extras := `Int (int_of_string flds.(!i)) :: !extras
+              incr i
+              extras := `Int (int_of_string flds.(!i)) :: !extras
+            (* strand *)
+            if cfg.frames = 6 then
+              incr i
+              extras := `String flds.(!i) :: !extras
+            (* other goodies *)
+            ([cfg.bls, `Float; cfg.anc_comp, `Float; cfg.dna, `String; cfg.aa, `String]) |> List.iter
+              fun (b,ty) ->
+                if b then
+                  incr i
+                  match ty with
+                    | `Float -> extras := `Float (float_of_string flds.(!i)) :: !extras
+                    | `String -> extras := `String flds.(!i) :: !extras
+            (* make output GTable row *)
+            Some (Array.of_list (`String alignment_name :: `Float sc :: (List.rev !extras)))
+
+(* main job: instantiate the job tree *)
+let main input =
+  let alignments_table = GTable.bind_link (input$"alignments")
+  let alignments_desc = GTable.describe alignments_table
+
+  (* resolve which parameter set to use *)
+  let species_set =
+    if input$?"species_set" then J.string (input$"species_set")
+    else
+      let alignments_deets = GTable.get_details alignments_table
+      let alignments_species =
+        try
+          J.string (alignments_deets$"maf_stitcher_input"$"species_set")
+        with _ ->
+          raise (AppError "Please provide the species_set input (could not detect which PhyloCSF parameters to use for these alignments)")
+      match alignments_species with
+        | "hg19_29mammals" -> "29mammals"
+        | "dm3_12flies" -> "12flies"
+        | _ -> raise (AppError (sprintf "No PhyloCSF parameters available for maf_stitcher species set: %s. Regenerate the alignments with a supported species set, or override this check by providing the species_set input" alignments_species))
+
+  (* verify the species names can be extracted from the alignments table.
+     TODO: also verify that the obtained species names match those supported by species_set *)
+  ignore (get_species_names alignments_table)
+
+  (* get own app[let] ID, for details breadcrumbs *)
+  let job_desc = api_call [job_id (); "describe"] J.empty
+  let app_id =
+    if job_desc$?"app" then J.string (job_desc$"app")
+    else J.string (job_desc$"applet")
+
+  (* initialize output GTable *)
+  let output_name =
+    if input$?"output_name" then J.string (input$"output_name")
+    else J.string (alignments_desc$"name") ^ " PhyloCSF"
+  let output_gtable = make_scores_gtable output_name app_id input
+
+  (* partition alignments and launch subjobs *)
+  let options =
+    if not (input$?"instance_type") then JSON.empty
+    else
+      J.of_assoc [
+        "systemRequirements", J.of_assoc [
+          "process", J.of_assoc [
+            "instanceType", (input$"instance_type")
+          ]
+        ]
+      ]
+  let subjobs =
+    partition_rows (JSON.int (alignments_desc$"length")) (JSON.int (input$"alignments_per_job")) |> List.map
+      fun (starting, limit) ->
+        let subjob = J.of_assoc [
+          "species_set", `String species_set;
+          "starting", `Int starting;
+          "limit", `Int limit;
+          "scores", `String (GTable.id output_gtable)
+        ]
+        new_job ~options "process" (input $+ ("process",subjob))
+
+  (* set up postprocessing job and output *)
+  let postprocess_input =
+    (input
+      $+ ("postprocess", J.of_assoc [
+            "scores", `String (GTable.id output_gtable);
+            "errors", J.of_list (List.map (fun subjob -> make_jobref subjob "errors") subjobs)]))
+  let postprocess = new_job "postprocess" postprocess_input
+
+  J.of_assoc [
+    "scores", make_link (GTable.id output_gtable);
+    "errors", make_jobref postprocess "errors"
+  ]
+
+(* subjob: process some of the alignments *)
+let process input =
+  let alignments_table = GTable.bind_link (input$"alignments")
+  let species_names = get_species_names alignments_table
+  let starting = J.int (input$"process"$"starting")
+  let limit = J.int (input$"process"$"limit")
+  let scores_table = GTable.bind (None,J.string (input$"process"$"scores"))
+
+  (* get configuration options *)
+  let cfg = {
+    strategy = J.string (input$"strategy");
+    frames = J.int (input$"frames");
+    orf = J.string (input$"orf");
+    min_codons = J.int (input$"min_codons");
+    all_scores = J.bool (input$"all_scores");
+    bls = J.bool (input$"bls");
+    anc_comp = J.bool (input$"anc_comp");
+    dna = J.bool (input$"dna");
+    aa = J.bool (input$"aa")
+  }
+
+  let worker_process alignment =
+    try
+      run_PhyloCSF (J.string (input$"process"$"species_set")) species_names alignment cfg
+    with
+      | DNAnexus.AppError msg ->
+          raise
+            ForkWork.ChildExn [
+              "AppError"; J.string ((GTable.json_of_row alignment)$@1); msg
+            ]
+      | exn ->
+          raise
+            ForkWork.ChildExn [
+              "AppInternalError";
+              J.string ((GTable.json_of_row alignment)$@1);
+              Printexc.to_string exn;
+              Printexc.get_backtrace ()
+            ]
+
+  let fw = ForkWork.manager ()
+  let rows = GTable.iterate_rows ~starting ~limit alignments_table
+  let promises = rows /@ (ForkWork.fork fw worker_process)
+  Enum.force promises
+  let errors =
+    promises |> Enum.fold
+      fun n ans ->
+        match ForkWork.await_result fw ans with
+          | `OK [] -> n+1
+          | `OK scores ->
+            scores |> List.iter (fun score_row -> GTable.add_row scores_table score_row)
+            n
+          | `Exn ["AppError"; name; msg] ->
+              eprintf "AppError on %s: %s\n" name msg
+              raise (DNAnexus.AppError (sprintf "%s: %s" name msg))
+          | `Exn ["AppInternalError"; name; msg; bt] ->
+              eprintf "AppInternalError on %s: %s\n" name msg
+              if bt <> "" then eprintf "%s\n" bt
+              raise (DNAnexus.AppInternalError (sprintf "%s: %s" name msg))
+          | _ -> assert false
+      0
+
+  GTable.flush_rows scores_table
+  J.of_assoc ["errors", `Int errors]
+
+(* finish-up job: collect stats & initiate GTable closure *)
+let postprocess input =
+  let gtable = GTable.bind (None, JSON.string (input$"postprocess"$"scores"))
+
+  let stats =
+    Enum.fold
+      fun failures j -> failures + JSON.int j
+      0
+      Vect.enum (JSON.array (input$"postprocess"$"errors"))
+  let stats_json = J.of_assoc ["errors", `Int stats]
+
+  GTable.set_details gtable (GTable.get_details gtable $+ ("PhyloCSF_stats", stats_json))
+
+  GTable.close gtable
+  J.of_assoc ["scores", make_link (GTable.id gtable); "errors", `Int stats]
+
+(* entry point *)
+job_main
+  fun input ->
+    if input $? "process" then process input
+    else if input $? "postprocess" then postprocess input
+    else main input
diff --git a/DNAnexus/src/run.sh b/DNAnexus/src/run.sh
new file mode 100755
index 0000000..536016e
--- /dev/null
+++ b/DNAnexus/src/run.sh
@@ -0,0 +1,16 @@
+#!/bin/bash
+
+# This script simply launches your C++ executable and contains 3 entry
+# points.
+
+main() {
+        /dxPhyloCSF
+}
+
+process() {
+        /dxPhyloCSF process
+}
+
+postprocess() {
+        /dxPhyloCSF postprocess
+}
diff --git a/DNAnexus/test/test.sh b/DNAnexus/test/test.sh
new file mode 100755
index 0000000..b16d11b
--- /dev/null
+++ b/DNAnexus/test/test.sh
@@ -0,0 +1,64 @@
+#!/bin/bash -e
+
+SOURCE="${BASH_SOURCE[0]}"
+while [ -h "$SOURCE" ] ; do SOURCE="$(readlink "$SOURCE")"; done
+HERE="$( cd -P "$( dirname "$SOURCE" )" && pwd )"
+
+function test_in_project {
+  dx-build-applet $HERE/.. -d $TEST_PROJECT:/
+  dx="dx --project-context-id=$TEST_PROJECT"
+  job1=$($dx run :/PhyloCSF -i alignments="PhyloCSF development:/test data/tal-AA dm3_12flies alignments" \
+           --delay-workspace-destruction -y --brief)
+  job2=$($dx run :/PhyloCSF -i alignments="PhyloCSF development:/test data/tal-AA dm3_12flies alignments" \
+           -i bls=true -i dna=true \
+           --delay-workspace-destruction -y --brief)
+  job3=$($dx run :/PhyloCSF -i alignments="PhyloCSF development:/test data/tal-AA dm3_12flies alignments" \
+           -i frames=6 -i all_scores=true -i bls=true -i anc_comp=true -i dna=true -i aa=true \
+           --delay-workspace-destruction -y --brief)
+  job4=$($dx run :/PhyloCSF -i alignments="PhyloCSF development:/test data/ALDH2_exons hg19_29mammals alignments" \
+           -i strategy=fixed -i orf=ATGStop -i frames=6 -i bls=true -i anc_comp=true -i aa=true \
+           --delay-workspace-destruction -y --brief)
+  job5=$($dx run :/PhyloCSF \
+           -i alignments="PhyloCSF development:/test data/CCDS longest isoforms hg19_29mammals chr21 alignments" \
+           -i strategy=fixed -i alignments_per_job=10 \
+           --delay-workspace-destruction -y --brief)
+
+  $dx wait $job1
+  $dx export tsv -o - $job1:scores
+  $dx wait $job2
+  $dx export tsv -o - $job2:scores
+  $dx wait $job3
+  $dx export tsv -o - $job3:scores
+  $dx wait $job4
+  $dx export tsv -o - $job4:scores
+  $dx wait $job5
+  printf "%d/197\n" $($dx export tsv -o - --no-header $job5:scores | cut -f2 | grep -v - | wc -l)
+}
+
+if [ "$1" = "inner" ]; then
+  test_in_project
+  exit
+fi
+
+TEST_PROJECT=$(dx new project --brief "PhyloCSF test ($(date))")
+export TEST_PROJECT
+
+function cleanup {
+  dx rmproject -y $TEST_PROJECT > /dev/null
+}
+trap "cleanup; exit" int
+
+$SOURCE inner &
+inner=$!
+
+exit_code=0
+wait $inner || exit_code=$?
+
+if [ "$exit_code" -ne 0 ]; then
+  echo "Leaving behind project for failed test, named \"$(dx describe $TEST_PROJECT --name)\""
+  echo "To delete: dx rmproject -y $TEST_PROJECT"
+else
+  cleanup
+fi
+
+exit $exit_code