From b029bff0ec25ec06758a5fdf7c79152d2ed5c032 Mon Sep 17 00:00:00 2001
From: Justin Bedo <cu@cua0.org>
Date: Fri, 4 Jan 2019 11:37:38 +1100
Subject: snpeff: init

---
 default.nix               |   1 +
 lib/references.nix        | 162 +++++++++++++++++++++++++++-------------------
 tools/snpeff-annotate.nix |  21 ++++++
 tools/snpeff-dbnsfp.nix   |  22 +++++++
 tools/snpeff.nix          |   8 +++
 5 files changed, 146 insertions(+), 68 deletions(-)
 create mode 100644 tools/snpeff-annotate.nix
 create mode 100644 tools/snpeff-dbnsfp.nix
 create mode 100644 tools/snpeff.nix

diff --git a/default.nix b/default.nix
index 27afcb5..538ddf4 100644
--- a/default.nix
+++ b/default.nix
@@ -26,6 +26,7 @@ let
     platypus = callBionix ./tools/platypus.nix {};
     samtools = callBionix ./tools/samtools.nix {};
     strelka = callBionix ./tools/strelka.nix {};
+    snpeff = callBionix ./tools/snpeff.nix {};
 
     ref = callBionix ./lib/references.nix {};
 
diff --git a/lib/references.nix b/lib/references.nix
index 1ac92de..9d1b947 100755
--- a/lib/references.nix
+++ b/lib/references.nix
@@ -33,90 +33,116 @@ rec {
       };
       buildInputs = [ gawk ];
       buildCommand = ''
-        gunzip < $src | awk '/^[^#]/{print "chr" $0;next}{print}' > $out
-      '';
-      passthru.filetype = filetype.vcf { ref = seq; };
-    };
-    cosmic = {coding, noncoding}: stdenvNoCC.mkDerivation rec {
-      name = "cosmic-grch38";
-      buildInputs = [ gawk ];
-      buildCommand = ''
+      gunzip < $src | awk '/^[^#]/{print "chr" $0;next}{print}' > $out
+        '';
+        passthru.filetype = filetype.vcf { ref = seq; };
+      };
+      cosmic = {coding, noncoding}: stdenvNoCC.mkDerivation rec {
+        name = "cosmic-grch38";
+        buildInputs = [ gawk ];
+        buildCommand = ''
         gunzip < ${coding} | grep '^#' > $out
         cat ${coding} ${noncoding} | gunzip | grep -v '^#' | sed 's/^/chr/' | sort -t$'\t' -k1,1 -k2,2n >> $out
-      '';
-      passthru.filetype = filetype.vcf { ref = seq; };
-    };
-    ensembl = {
-      cdna = stdenvNoCC.mkDerivation rec {
-        name = "ensembl-grch38-cdna-${version}";
-        version = "94";
-        src = fetchurl {
-          url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh38.cdna.all.fa.gz";
-          sha256 = "1fc5d6p2wlwsm49wnmxmm3byjx5jvr6z9fpzrq7v7fpb086adl0h";
+        '';
+        passthru.filetype = filetype.vcf { ref = seq; };
+      };
+      ensembl = {
+        cdna = stdenvNoCC.mkDerivation rec {
+          name = "ensembl-grch38-cdna-${version}";
+          version = "94";
+          src = fetchurl {
+            url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/cdna/Homo_sapiens.GRCh38.cdna.all.fa.gz";
+            sha256 = "1fc5d6p2wlwsm49wnmxmm3byjx5jvr6z9fpzrq7v7fpb086adl0h";
+          };
+          buildCommand = "gunzip < $src > $out";
+          passthru.filetype = filetype.fa {};
+        };
+        ncrna = stdenvNoCC.mkDerivation rec {
+          name = "ensembl-grch38-ncrna-${version}";
+          version = "94";
+          src = fetchurl {
+            url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/ncrna/Homo_sapiens.GRCh38.ncrna.fa.gz";
+            sha256 = "1cpasykwriila52nqgvw6d3mjyh6d9qi613hvhn4h1dxkqzgnjff";
+          };
+          buildCommand = "gunzip < $src > $out";
+          passthru.filetype = filetype.fa {};
         };
-        buildCommand = "gunzip < $src > $out";
-        passthru.filetype = filetype.fa {};
       };
-      ncrna = stdenvNoCC.mkDerivation rec {
-        name = "ensembl-grch38-ncrna-${version}";
-        version = "94";
-        src = fetchurl {
-          url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/homo_sapiens/ncrna/Homo_sapiens.GRCh38.ncrna.fa.gz";
-          sha256 = "1cpasykwriila52nqgvw6d3mjyh6d9qi613hvhn4h1dxkqzgnjff";
+      snpeff = {
+        db = stdenv.mkDerivation rec {
+          name = "GRCh38.86";
+          src = fetchurl {
+            url = "mirror://sourceforge/project/snpeff/databases/v4_3/snpEff_v4_3_${name}.zip";
+            sha256 = "1rf8q7l732ayjq2lpny4s75zpij05j00151374nqblk4wri2mz0i";
+          };
+          buildInputs = [ unzip ];
+          buildCommand = ''
+            unzip ${src}
+            mv data/${name} $out
+          '';
+        };
+        dbnsfp = {
+          db = requireFile {
+            name = "dbNSFP.txt.gz";
+            message = "download the dbNSFP database manually from https://drive.google.com/uc?export=download&id=0B7Ms5xMSFMYlbTZodjlGUDZnTGc and add to nix store";
+            sha256 = "0gahnwkc7v2q6p6ixkhvsgqvvm6xf0c3bdh4nf0alih83h3wffd0";
+          };
+          index = requireFile {
+            name = "dbNSFP.txt.gz.tbi";
+            message = "download the dbNSFP index manually from https://drive.google.com/uc?export=download&id=0B7Ms5xMSFMYlNVBJdFA5cFZRYkE and add to nix store";
+            sha256 = "18blkly6gvg7r0sx968xlb1zl2kqg5j1kpbrm2r7ajlxlfyvrx3w";
+          };
         };
-        buildCommand = "gunzip < $src > $out";
-        passthru.filetype = filetype.fa {};
       };
     };
-  };
 
-  grcm38 = grcm38-p6;
-  grcm38-p6 = {
-    seq = stdenvNoCC.mkDerivation rec {
-      name = "seq-grcm38.${version}";
-      version = "p6";
-      src = fetchurl {
-        url = "ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_mouse/release_M19/GRCm38.${version}.genome.fa.gz";
-        sha256 = "0ryiqab5bldpzawylsk2qpjxr2j701q03ww9jqyxhkimqpn9g3mr";
-      };
-      buildCommand = "gunzip < $src > $out";
-      passthru.filetype = filetype.fa {};
-    };
-    ensembl = {
-      cdna = stdenvNoCC.mkDerivation rec {
-        name = "ensembl-grch38-cdna-${version}";
-        version = "94";
+    grcm38 = grcm38-p6;
+    grcm38-p6 = {
+      seq = stdenvNoCC.mkDerivation rec {
+        name = "seq-grcm38.${version}";
+        version = "p6";
         src = fetchurl {
-          url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/mus_musculus/cdna/Mus_musculus.GRCm38.cdna.all.fa.gz";
-          sha256 = "0khp9l6s35lav2xqp7vkk6ybnz4wjihn7lapjf2lbpnbzjb4hp6d";
+          url = "ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_mouse/release_M19/GRCm38.${version}.genome.fa.gz";
+          sha256 = "0ryiqab5bldpzawylsk2qpjxr2j701q03ww9jqyxhkimqpn9g3mr";
         };
         buildCommand = "gunzip < $src > $out";
         passthru.filetype = filetype.fa {};
       };
-      ncrna = stdenvNoCC.mkDerivation rec {
-        name = "ensembl-grch38-ncrna-${version}";
-        version = "94";
-        src = fetchurl {
-          url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/mus_musculus/ncrna/Mus_musculus.GRCm38.ncrna.fa.gz";
-          sha256 = "0d997gm8p2b89rm5d46m2x4vz9lijxarfr2lzylnbi8gyqrbagdd";
+      ensembl = {
+        cdna = stdenvNoCC.mkDerivation rec {
+          name = "ensembl-grch38-cdna-${version}";
+          version = "94";
+          src = fetchurl {
+            url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/mus_musculus/cdna/Mus_musculus.GRCm38.cdna.all.fa.gz";
+            sha256 = "0khp9l6s35lav2xqp7vkk6ybnz4wjihn7lapjf2lbpnbzjb4hp6d";
+          };
+          buildCommand = "gunzip < $src > $out";
+          passthru.filetype = filetype.fa {};
+        };
+        ncrna = stdenvNoCC.mkDerivation rec {
+          name = "ensembl-grch38-ncrna-${version}";
+          version = "94";
+          src = fetchurl {
+            url = "ftp://ftp.ensembl.org/pub/release-${version}/fasta/mus_musculus/ncrna/Mus_musculus.GRCm38.ncrna.fa.gz";
+            sha256 = "0d997gm8p2b89rm5d46m2x4vz9lijxarfr2lzylnbi8gyqrbagdd";
+          };
+          buildCommand = "gunzip < $src > $out";
+          passthru.filetype = filetype.fa {};
         };
-        buildCommand = "gunzip < $src > $out";
-        passthru.filetype = filetype.fa {};
       };
     };
-  };
 
-  mm10 = mm10-p4;
-  mm10-p4 = {
+    mm10 = mm10-p4;
+    mm10-p4 = {
       seq = stdenvNoCC.mkDerivation rec {
-          name = "seq-mm10.${version}";
-          version = "p4";
-          src = fetchurl {
-              url = "http://hgdownload.soe.ucsc.edu/goldenPath/mm10/mm10Patch4/mm10Patch4.fa.gz";
-              sha256 = "1660d6d05f3aa266c6053cfd1efef1747d9e854836917241d6f47cff7a55340c";
-              };
-          buildCommand = "gunzip < $src > $out";
-          passthru.filetype = filetype.fa {};
+        name = "seq-mm10.${version}";
+        version = "p4";
+        src = fetchurl {
+          url = "http://hgdownload.soe.ucsc.edu/goldenPath/mm10/mm10Patch4/mm10Patch4.fa.gz";
+          sha256 = "1660d6d05f3aa266c6053cfd1efef1747d9e854836917241d6f47cff7a55340c";
+        };
+        buildCommand = "gunzip < $src > $out";
+        passthru.filetype = filetype.fa {};
       };
-  };
-}
+    };
+  }
diff --git a/tools/snpeff-annotate.nix b/tools/snpeff-annotate.nix
new file mode 100644
index 0000000..a119c3c
--- /dev/null
+++ b/tools/snpeff-annotate.nix
@@ -0,0 +1,21 @@
+{bionix
+,nixpkgs
+,db
+,flags ? ""}:
+
+input:
+
+with nixpkgs;
+with bionix.types;
+
+assert (matchFiletype "snpeff-annotate" { vcf = _: true; } input);
+
+stdenv.mkDerivation {
+  name = "snpeff-annotate";
+  buildCommand = ''
+    ln -s ${db} ${db.name}
+    snpeff -nodownload -dataDir $TMPDIR ${db.name} ${input} > $out
+  '';
+  buildInputs = [ snpeff ];
+  passthru.filetype = input.filetype;
+}
diff --git a/tools/snpeff-dbnsfp.nix b/tools/snpeff-dbnsfp.nix
new file mode 100644
index 0000000..f9fbdaa
--- /dev/null
+++ b/tools/snpeff-dbnsfp.nix
@@ -0,0 +1,22 @@
+{bionix
+,nixpkgs
+,dbnsfp
+,flags ? ""}:
+
+input:
+
+with nixpkgs;
+with bionix.types;
+
+assert (matchFiletype "snpeff-dbnsfp" { vcf = _: true; } input);
+
+stdenv.mkDerivation {
+  name = "snpeff-dbnsfp";
+  buildCommand = ''
+    ln -s ${dbnsfp.db} dbNSFP.txt.gz
+    ln -s ${dbnsfp.index} dbNSFP.txt.gz.tbi
+    snpeff dbnsfp -db dbNSFP.txt.gz ${input} > $out
+  '';
+  buildInputs = [ snpeff ];
+  passthru.filetype = input.filetype;
+}
diff --git a/tools/snpeff.nix b/tools/snpeff.nix
new file mode 100644
index 0000000..2df8056
--- /dev/null
+++ b/tools/snpeff.nix
@@ -0,0 +1,8 @@
+{bionix, nixpkgs}:
+
+with bionix;
+
+{
+  annotate = callBionixE ./snpeff-annotate.nix;
+  dbnsfp = callBionixE ./snpeff-dbnsfp.nix;
+}
-- 
cgit v1.2.3