From 5cb6f6001ea3aefd52ebaf42c1714343d51a8822 Mon Sep 17 00:00:00 2001
From: Justin Bedo <cu@cua0.org>
Date: Thu, 10 Jan 2019 11:47:05 +1100
Subject: cnvkit: init

---
 default.nix            |   6 +--
 tools/cnvkit-app.nix   | 102 +++++++++++++++++++++++++++++++++++++++++++++++++
 tools/cnvkit-batch.nix |  41 ++++++++++++++++++++
 tools/cnvkit.nix       |   8 ++++
 4 files changed, 154 insertions(+), 3 deletions(-)
 create mode 100644 tools/cnvkit-app.nix
 create mode 100644 tools/cnvkit-batch.nix
 create mode 100644 tools/cnvkit.nix

diff --git a/default.nix b/default.nix
index c50e063..3ed2b50 100644
--- a/default.nix
+++ b/default.nix
@@ -13,8 +13,9 @@ let
 
     types = callBionix ./lib/types.nix {};
 
-    bwa = callBionix ./tools/bwa.nix {};
     bowtie = callBionix ./tools/bowtie.nix {};
+    bwa = callBionix ./tools/bwa.nix {};
+    cnvkit = callBionix ./tools/cnvkit.nix {};
     compression = callBionix ./tools/compression.nix {};
     crumble = callBionix ./tools/crumble.nix {};
     fastqc = callBionix ./tools/fastqc.nix {};
@@ -25,9 +26,8 @@ let
     mutect = callBionix ./tools/mutect.nix {};
     platypus = callBionix ./tools/platypus.nix {};
     samtools = callBionix ./tools/samtools.nix {};
-    strelka = callBionix ./tools/strelka.nix {};
     snpeff = callBionix ./tools/snpeff.nix {};
-
+    strelka = callBionix ./tools/strelka.nix {};
     ref = callBionix ./lib/references.nix {};
 
     qsub = attrs: bionix.extend (self: super: with self; rec {
diff --git a/tools/cnvkit-app.nix b/tools/cnvkit-app.nix
new file mode 100644
index 0000000..d44c508
--- /dev/null
+++ b/tools/cnvkit-app.nix
@@ -0,0 +1,102 @@
+{ lib
+, fetchurl
+, rPackages
+, rWrapper
+, buildPythonPackage
+, fetchPypi
+, biopython
+, six
+, numpy
+, scipy
+, pandas
+, matplotlib
+, reportlab
+, zlib
+, bzip2
+, htslib
+, libjpeg
+, pkgconfig
+, futures
+}:
+
+let pyfaidx = buildPythonPackage rec {
+      pname = "pyfaidx";
+      version = "0.5.3.1";
+
+      src = fetchPypi {
+        inherit pname version;
+        sha256 = "0mjbksbj9hh2cf0yjr951cjahhn0lg7p71kd3kvbnscqyxa44kfr";
+      };
+
+      propagatedBuildInputs = [ six ];
+    };
+
+    pysam = buildPythonPackage rec {
+      pname = "pysam";
+      version = "0.14";
+
+      src = fetchPypi {
+        inherit pname version;
+        sha256 = "11snrwl5pn6knarf0c7zcpkhc7rijb6cfpw8sl3pdvnynj1fmj69";
+      };
+
+      buildInputs = [ zlib ];
+
+      preConfigure = ''
+        export HTSLIB_MODE="external"
+        export HTSLIB_LIBRARY_DIR=${htslib}/lib
+        export HTSLIB_INCLUDE_DIR=${htslib}/include
+      '';
+
+    };
+
+    future = buildPythonPackage rec {
+      pname = "future";
+      version = "0.16.0";
+
+      doCheck = false;
+
+      src = fetchPypi {
+        inherit pname version;
+        sha256 = "1nzy1k4m9966sikp0qka7lirh8sqrsyainyf8rk97db7nwdfv773";
+      };
+    };
+
+    pillow = buildPythonPackage rec {
+      pname = "Pillow";
+      version = "5.0.0";
+
+      doCheck = false;
+
+      src = fetchPypi {
+        inherit pname version;
+        sha256 = "1fz1n1cq65dqdbiwy1cn8l21amqbzq18rdmmcr670ks24dn9vwhj";
+      };
+
+      buildInputs = [ zlib libjpeg pkgconfig ];
+    };
+
+    cghFLasso = rPackages.buildRPackage rec {
+      name = "cghFLasso-${version}";
+      version = "0.2-1";
+      src = fetchurl {
+        url = "https://cran.r-project.org/src/contrib/Archive/cghFLasso/cghFLasso_${version}.tar.gz";
+        sha256 = "0b1hnjf9g0v47hbz0dy9m6jhcl1ky20yyhhmm8myng2sndcpjsbf";
+      };
+    };
+
+    cnvR = rWrapper.override {
+      packages = with rPackages; [ DNAcopy cghFLasso ];
+    };
+
+in buildPythonPackage rec {
+  pname = "CNVkit";
+  version = "0.9.5";
+
+  src = fetchPypi {
+    inherit pname version;
+    sha256 = "1sa70bmnxj1lzp33pbj3axk6n77czswwj9cirimxh2qrn84i7vs3";
+  };
+
+  propagatedBuildInputs = [ biopython numpy scipy pandas matplotlib reportlab pyfaidx pysam futures future pillow cnvR ];
+}
diff --git a/tools/cnvkit-batch.nix b/tools/cnvkit-batch.nix
new file mode 100644
index 0000000..6b69f1a
--- /dev/null
+++ b/tools/cnvkit-batch.nix
@@ -0,0 +1,41 @@
+{bionix
+,normals ? []
+,targets ? null
+,annotations ? null
+,flags ? null
+,indexAttrs ? {}}:
+
+{normals ? normals, tumours}:
+
+with bionix;
+with lib;
+with types;
+
+let
+  getref = f: matchFiletype "cnvkit-batch" { bam = {ref, ...}: ref; } f;
+  refs = map getref normals ++ map getref tumours;
+  ref = head refs;
+  sorted = matchFileSorting "cnvkit-batch" { coord = _: true; };
+in
+
+assert (length (unique refs) == 1);
+assert (all sorted (normals ++ tumours));
+
+stage {
+  name = "cnvkit";
+  buildInputs = [ cnvkit.app ];
+  buildCommand = ''
+    ln -s ${ref} ref.fa
+    ln -s ${samtools.faidx indexAttrs ref} ref.fa.fai
+    cnvkit.py batch ${concatStringsSep " " tumours} \
+      ${optionalString (normals != []) ("-n " + concatStringsSep " " normals)} \
+      ${optionalString (annotations != null) annotations} \
+      ${if targets != null then "--targets ${targets}" else "-m wgs"} \
+      -f ref.fa \
+      -p $NIX_BUILD_CORES \
+      -d $TMPDIR \
+      ${optionalString (flags != null) flags}
+    mkdir $out
+    cp * $out
+  '';
+}
diff --git a/tools/cnvkit.nix b/tools/cnvkit.nix
new file mode 100644
index 0000000..610b18c
--- /dev/null
+++ b/tools/cnvkit.nix
@@ -0,0 +1,8 @@
+{bionix}:
+
+with bionix;
+
+{
+app = lib.callPackageWith (pkgs // pkgs.pythonPackages) ./cnvkit-app.nix {};
+  callCNV = callBionixE ./cnvkit-batch.nix;
+}
-- 
cgit v1.2.3