hisat2: add version 2.2.1 and sra variant (#34966)

- add python dependency
- add perl dependency
- set python and perl interpreters in scripts

2 files changed, 77 insertions, 4 deletions

diff --git a/var/spack/repos/builtin/packages/hisat2/package.py b/var/spack/repos/builtin/packages/hisat2/package.py
index 2542942a21..37dd78b60e 100644
--- a/var/spack/repos/builtin/packages/hisat2/package.py
+++ b/var/spack/repos/builtin/packages/hisat2/package.py
@@ -3,6 +3,7 @@
 #
 # SPDX-License-Identifier: (Apache-2.0 OR MIT)
 
+import glob
 
 from spack.package import *
 
@@ -17,6 +18,12 @@ class Hisat2(MakefilePackage):
     url = "ftp://ftp.ccb.jhu.edu/pub/infphilo/hisat2/downloads/hisat2-2.1.0-source.zip"
 
     version(
+        "2.2.1",
+        sha256="48e933330d4d8470d2b3dfe7ec3918f2e98a75f7381891e23b7df1fb4f135eb1",
+        url="https://cloud.biohpc.swmed.edu/index.php/s/fE9QCsX3NH4QwBi/download",
+        extension="zip",
+    )
+    version(
         "2.2.0",
         sha256="0dd55168853b82c1b085f79ed793dd029db163773f52272d7eb51b3b5e4a4cdd",
         url="https://cloud.biohpc.swmed.edu/index.php/s/hisat2-220-source/download",
@@ -24,14 +31,36 @@ class Hisat2(MakefilePackage):
     )
     version("2.1.0", sha256="89a276eed1fc07414b1601947bc9466bdeb50e8f148ad42074186fe39a1ee781")
 
+    variant("sra", default=False, description="Add SRA (Sequence Read Archive) support")
+
+    depends_on("perl", type="run")
+    depends_on("python", type="run")
+
+    depends_on("sra-tools", when="+sra")
+    depends_on("ncbi-vdb", when="+sra")
+
+    # patch to get SRA working
+    patch("sra.patch", when="+sra")
+
+    @when("+sra")
+    def build(self, spec, prefix):
+        make(
+            "USE_SRA=1",
+            "NCBI_NGS_DIR={0}".format(spec["sra-tools"].prefix),
+            "NCBI_VDB_DIR={0}".format(spec["ncbi-vdb"].prefix),
+        )
+
     def install(self, spec, prefix):
         if spec.satisfies("@:2.1.0"):
             install_tree("doc", prefix.doc)
 
         install_tree("example", prefix.example)
-        install_tree("hisatgenotype_modules", prefix.hisatgenotype_modules)
-        install_tree("hisatgenotype_scripts", prefix.hisatgenotype_scripts)
         install_tree("scripts", prefix.scripts)
+
+        if "@:2.2.0" in spec:
+            install_tree("hisatgenotype_modules", prefix.hisatgenotype_modules)
+            install_tree("hisatgenotype_scripts", prefix.hisatgenotype_scripts)
+
         mkdirp(prefix.bin)
         install("hisat2", prefix.bin)
         install("hisat2-align-s", prefix.bin)
@@ -44,5 +73,38 @@ class Hisat2(MakefilePackage):
         install("hisat2-inspect-l", prefix.bin)
         install("*.py", prefix.bin)
 
-    def setup_run_environment(self, env):
-        env.prepend_path("PATH", self.spec.prefix)
+        if "@2.2:" in spec:
+            install("hisat2-repeat", prefix.bin)
+
+    @run_after("install")
+    def filter_sbang(self):
+        with working_dir(self.prefix.bin):
+            pattern = "^#!.*/usr/bin/env python"
+            repl = "#!{0}".format(self.spec["python"].command.path)
+            files = [
+                "hisat2-build",
+                "hisat2-inspect",
+            ]
+            for file in files:
+                filter_file(pattern, repl, *files, backup=False)
+
+            pattern = "^#!.*/usr/bin/env perl"
+            repl = "#!{0}".format(self.spec["perl"].command.path)
+            files = [
+                "hisat2",
+            ]
+            for file in files:
+                filter_file(pattern, repl, *files, backup=False)
+
+            pattern = "^#!.*/usr/bin/env python3"
+            repl = "#!{0}".format(self.spec["python"].command.path)
+            files = glob.glob("*.py")
+            for file in files:
+                filter_file(pattern, repl, *files, backup=False)
+
+        with working_dir(self.prefix.scripts):
+            pattern = "^#!.*/usr/bin/perl"
+            repl = "#!{0}".format(self.spec["perl"].command.path)
+            files = glob.glob("*.pl")
+            for file in files:
+                filter_file(pattern, repl, *files, backup=False)
diff --git a/var/spack/repos/builtin/packages/hisat2/sra.patch b/var/spack/repos/builtin/packages/hisat2/sra.patch
new file mode 100644
index 0000000000..b7cbec035c
--- /dev/null
+++ b/var/spack/repos/builtin/packages/hisat2/sra.patch
@@ -0,0 +1,11 @@
+--- a/Makefile	2020-07-24 10:46:16.000000000 -0500
++++ b/Makefile	2023-01-12 20:28:08.919588735 -0600
+@@ -94,7 +94,7 @@
+ SERACH_INC = 
+ ifeq (1,$(USE_SRA))
+ 	SRA_DEF = -DUSE_SRA
+-	SRA_LIB = -lncbi-ngs-c++-static -lngs-c++-static -lncbi-vdb-static -ldl
++	SRA_LIB = -lncbi-ngs-c++-static -lngs-c++-static -lncbi-ngs-static -lncbi-vdb-static -ldl
+ 	SEARCH_INC += -I$(NCBI_NGS_DIR)/include -I$(NCBI_VDB_DIR)/include
+ 	SEARCH_LIBS += -L$(NCBI_NGS_DIR)/lib64 -L$(NCBI_VDB_DIR)/lib64
+ endif