Add lua for tensorqtl 1.0.8

tensorqtl/1.0.8.lua

@@ -0,0 +1,32 @@
+help([[
+This module loads tensorQTL 1.0.8 for python 3.9.0 packaged with conda_R/4.3.
+]])
+
+if not string.match(os.getenv("HOSTNAME"),"compute") and not  string.match(os.getenv("HOSTNAME"),"transfer") then  
+    LmodError("\
+This package can only be loaded on a compute or transfer node. Please use qrsh to connect to a valid compute or transfer node.")
+end
+
+if (mode() == "load") then
+    LmodMessage("Loading LIBD module for tensorQTL/1.0.8")
+elseif (mode() == "unload") then
+    LmodMessage("Unloading LIBD module for tensorQTL/1.0.8")
+end
+
+load('conda_R/4.3')
+
+-- Directly activate or deactivate the python virtual environment
+execute {cmd="source /jhpce/shared/jhpce/libd/tensorqtl/1.0.8/tensorqtl_venv/bin/activate", modeA={"load"}}
+execute {cmd="deactivate", modeA={"unload"}}
+
+-- Seems to be needed for the 'rfunc' module in tensorqtl
+prepend_path("LD_LIBRARY_PATH", "/jhpce/shared/jhpce/core/conda/miniconda3-4.11.0/envs/svnR-4.3/R/4.3/lib64/R/lib")
+
+-- Since there is both a module and package/directory named "tensorqtl", (and
+-- other modules in the same directory as the "tensorqtl" module), construct the
+-- PYTHONPATH such that "import tensorqtl" and "from tensorqtl import *" behave
+-- as expected (refer to different "tensorqtl"s). This is necessary because the
+-- pip version of 1.0.8 hasn't been released, but the GitHub repo contains the
+-- up-to-date 1.0.8 python code
+prepend_path("PYTHONPATH", "/jhpce/shared/jhpce/libd/tensorqtl/1.0.8/tensorqtl/tensorqtl")
+prepend_path("PYTHONPATH", "/jhpce/shared/jhpce/libd/tensorqtl/1.0.8/tensorqtl")