Fix HPC modules

CHANGELOG.md

@@ -1,6 +1,7 @@
 ## Tools development version
 
 - use major & minor version for docs website subdirectories. (#15, @kelly-sovacool)
+- Fig bug where `nextflow.run()` did not import the correct HPC modules. (#20, @kelly-sovacool)
 
 ## Tools 0.1.1
 

scripts/which_vpn.sh

@@ -1,5 +1,6 @@
-#!/bin/bash
+#!/usr/bin/env bash
 # trying to find out which VPN you are connected to??
+set -euo pipefail
 
 if [[ "$HOSTNAME" == "biowulf.nih.gov" ]]
 then

src/ccbr_tools/pipeline/hpc.py

@@ -5,8 +5,8 @@
 which contains default attributes for supported clusters.
 """
 
-from .util import get_hpcname
 from .cache import get_singularity_cachedir
+from ..shell import shell_run
 
 
 class Cluster:
@@ -53,8 +53,16 @@
         super().__init__()
         self.name = "biowulf"
         self.modules = {
-            "nxf": "ccbrpipeliner nextflow",
-            "smk": "ccbrpipeliner snakemake/7 singularity",
+            "nxf": (
+                "nextflow" + ""
+                if is_loaded(module="ccbrpipeliner")
+                else " ccbrpipeliner"
+            ),
+            "smk": (
+                "snakemake/7 singularity" + ""
+                if is_loaded(module="ccbrpipeliner")
+                else " ccbrpipeliner"
+            ),
         }
         self.singularity_sif_dir = "/data/CCBR_Pipeliner/SIFs"
 
@@ -102,3 +110,47 @@
     hpc_options = {"biowulf": Biowulf, "frce": FRCE}
     hpc_name = get_hpcname() if not debug else debug
     return hpc_options.get(hpc_name, Cluster)()
+
+
+def get_hpcname():
+    """
+    Get the HPC name using scontrol
+
+    Returns:
+        hpcname (str): The HPC name  (biowulf, frce, or an empty string)
+    """
+    scontrol_out = scontrol_show()
+    hpc = scontrol_out["ClusterName"] if "ClusterName" in scontrol_out.keys() else ""
+    if hpc == "fnlcr":
+        hpc = "frce"
+    return hpc
+
+
+def scontrol_show():
+    """
+    Run `scontrol show config` and parse the output as a dictionary
+
+    Returns:
+        scontrol_dict (dict): dictionary containing the output of `scontrol show config`
+    """
+    scontrol_dict = dict()
+    scontrol_out = shell_run(
+        "scontrol show config", shell=True, capture_output=True, text=True, check=False
+    )
+    if len(scontrol_out) > 0:
+        for line in scontrol_out.split("\n"):
+            line_split = line.split("=")
+            if len(line_split) > 1:
+                scontrol_dict[line_split[0].strip()] = line_split[1].strip()
+    return scontrol_dict
+
+
+def is_loaded(module="ccbrpipeliner"):
+    """
+    Check whether the ccbrpipeliner module is loaded
+
+    Returns:
+        is_loaded (bool): True if the module is loaded, False otherwise
+    """
+    output = shell_run("bash -c 'module list'", check=False)
+    return module in output

src/ccbr_tools/pipeline/nextflow.py

@@ -19,6 +19,7 @@ def run(
     mode="local",
     pipeline_name=None,
     debug=False,
+    hpc=get_hpc(),
 ):
     """
     Run a Nextflow workflow
@@ -33,7 +34,6 @@ def run(
     """
     nextflow_command = ["nextflow", "run", nextfile_path]
 
-    hpc = get_hpc()
     if mode == "slurm" and not hpc:
         raise ValueError("mode is 'slurm' but no HPC environment was detected")
     # add any additional Nextflow commands
@@ -60,10 +60,8 @@ def run(
     ):  # only add to the profiles if there are any. there are none when champagne is run on GitHub Actions.
         args_dict["-profile"] = ",".join(sorted(profiles))
     nextflow_command += list(f"{k} {v}" for k, v in args_dict.items())
-
-    # Print nextflow command
+    # turn command into a string
     nextflow_command = " ".join(str(nf) for nf in nextflow_command)
-    msg_box("Nextflow command", errmsg=nextflow_command)
 
     if mode == "slurm":
         slurm_filename = "submit_slurm.sh"
@@ -83,11 +81,13 @@ def run(
         msg_box("Slurm batch job", errmsg=run_command)
     elif mode == "local":
         if hpc:
-            nextflow_command = f'bash -c "module load {hpc.modules} && {hpc.env_vars} && {nextflow_command}"'
+            hpc_modules = hpc.modules["nxf"]
+            nextflow_command = f'bash -c "module load {hpc_modules} && {hpc.env_vars} && {nextflow_command}"'
         run_command = nextflow_command
     else:
         raise ValueError(f"mode {mode} not recognized")
 
     # Run Nextflow
+    msg_box("Nextflow command", errmsg=nextflow_command)
     if not debug:
         shell_run(run_command, capture_output=False)

src/ccbr_tools/pipeline/util.py

@@ -17,39 +17,7 @@
 import yaml
 
 from ..pkg_util import repo_base, msg
-
-
-def scontrol_show():
-    """
-    Run `scontrol show config` and parse the output as a dictionary
-
-    Returns:
-        scontrol_dict (dict): dictionary containing the output of `scontrol show config`
-    """
-    scontrol_dict = dict()
-    scontrol_out = subprocess.run(
-        "scontrol show config", shell=True, capture_output=True, text=True
-    ).stdout
-    if len(scontrol_out) > 0:
-        for line in scontrol_out.split("\n"):
-            line_split = line.split("=")
-            if len(line_split) > 1:
-                scontrol_dict[line_split[0].strip()] = line_split[1].strip()
-    return scontrol_dict
-
-
-def get_hpcname():
-    """
-    Get the HPC name using scontrol
-
-    Returns:
-        hpcname (str): The HPC name  (biowulf, frce, or an empty string)
-    """
-    scontrol_out = scontrol_show()
-    hpc = scontrol_out["ClusterName"] if "ClusterName" in scontrol_out.keys() else ""
-    if hpc == "fnlcr":
-        hpc = "frce"
-    return hpc
+from .hpc import get_hpcname
 
 
 def get_tmp_dir(tmp_dir, outdir, hpc=get_hpcname()):

tests/test_cli.py

@@ -1,5 +1,5 @@
 from ccbr_tools.shell import shell_run
-from ccbr_tools.pipeline.util import get_hpcname
+from ccbr_tools.pipeline.hpc import get_hpcname
 
 
 def test_version():

tests/test_hpc.py

@@ -1,4 +1,6 @@
 from ccbr_tools.pipeline.hpc import get_hpc
+from ccbr_tools.shell import shell_run
+import subprocess
 
 
 def test_hpc_biowulf():
@@ -27,3 +29,7 @@ def test_hpc_frce():
 def test_hpc_none():
     hpc = get_hpc(debug=" ")
     assert not any([hpc, hpc.name, *hpc.modules.values(), hpc.singularity_sif_dir])
+
+
+if __name__ == "__main__":
+    print(get_hpc(debug="biowulf"))

tests/test_nextflow.py

@@ -1,8 +1,33 @@
 from ccbr_tools.pipeline.nextflow import run
+from ccbr_tools.pipeline.hpc import get_hpc, Biowulf, FRCE
 from ccbr_tools.shell import exec_in_context
 
+import tempfile
+
 
 def test_nextflow_basic():
-    assert "nextflow run CCBR/CHAMPAGNE" in exec_in_context(
-        run, "CCBR/CHAMPAGNE", debug=True
+    output = exec_in_context(run, "CCBR/CHAMPAGNE", debug=True)
+    assert "nextflow run CCBR/CHAMPAGNE" in output
+
+
+def test_nextflow_hpc():
+    assert all(
+        [
+            "module load"
+            in exec_in_context(run, "CCBR/CHAMPAGNE", debug=True, hpc=Biowulf()),
+            "module load"
+            in exec_in_context(run, "CCBR/CHAMPAGNE", debug=True, hpc=FRCE()),
+        ]
+    )
+
+
+def test_nextflow_slurm():
+    assert "sbatch submit_slurm.sh" in exec_in_context(
+        run, "CCBR/CHAMPAGNE", debug=True, hpc=Biowulf(), mode="slurm"
+    )
+
+
+if __name__ == "__main__":
+    print(
+        exec_in_context(run, "CCBR/CHAMPAGNE", debug=True, hpc=Biowulf(), mode="slurm")
     )

tests/test_pipeline_util.py

@@ -0,0 +1,11 @@
+from ccbr_tools.pipeline.util import get_tmp_dir
+
+
+def test_tmp_dir():
+    assert all(
+        [
+            get_tmp_dir("", "./out", hpc="biowulf").startswith("/lscratch"),
+            get_tmp_dir("", "./out", hpc="frce").startswith("./out"),
+            get_tmp_dir("", "./out", hpc="none") == None,
+        ]
+    )

tests/test_scripts.py

@@ -1,5 +1,6 @@
+import pytest
 from ccbr_tools.shell import shell_run
-from ccbr_tools.pipeline.util import get_hpcname
+from ccbr_tools.pipeline.hpc import get_hpcname
 
 
 def test_scripts_help():
@@ -9,7 +10,7 @@ def test_scripts_help():
 
 
 def test_which_vpn():
-    which_vpn = shell_run("which_vpn.sh")
+    which_vpn = shell_run("which_vpn.sh", check=False)
     hpc = get_hpcname()
     if hpc == "biowulf":
         assert "DO NOT RUN THIS ON a BIOWULF interactive node!" in which_vpn