Merge pull request #751 from haddocking/bugfix_traceback_contactmap

Bugfix traceback contactmap

examples/analysis/contmap-test.cfg

@@ -18,7 +18,7 @@ mode = "local"
 # Input molecules
 molecules = ["../data/1a2k_r_u.pdb", "../data/1a2k_l_u.pdb"]
 # Disable post-processing
-postprocess = false
+postprocess = true
 
 # =================== #
 # Workflow definition #

src/haddock/clis/cli_traceback.py

@@ -25,7 +25,35 @@
 
 TRACK_FOLDER = "traceback"  # name of the traceback folder
 
-ANA_MODULES = ["caprieval", "seletop", "topoaa", "rmsdmatrix", "clustrmsd", "clustfcc"]
+
+def get_steps_without_pdbs(run_dir, all_steps):
+    """
+    Get the modules that do not produce PDB files.
+
+    Parameters
+    ----------
+    run_dir : str or pathlib.Path
+        Path to the run directory.
+    
+    all_steps : list
+        List of all the steps in the run directory.
+    
+    Returns
+    -------
+    steps_without_pdbs : list
+        List of steps that did not produce PDB files.
+    """
+    steps_without_pdbs = []
+    for step in all_steps:
+        if step.endswith("topoaa"):
+            steps_without_pdbs.append(step)
+        else:
+            step_dir = Path(run_dir, step)
+            if step_dir.is_dir():
+                pdbs = list(step_dir.glob("*.pdb*"))
+                if len(pdbs) == 0:
+                    steps_without_pdbs.append(step)
+    return steps_without_pdbs
 
 
 def get_ori_names(n: int, pdbfile: PDBFile, max_topo_len: int) -> tuple[list, int]:
@@ -155,7 +183,9 @@ def main(run_dir):
     # get the module folders from the run_dir input
     all_steps = get_module_steps_folders(Path(run_dir))
     log.info(f"All_steps: {', '.join(all_steps)}")
-    sel_step = [st for st in all_steps if st.split("_")[1] not in ANA_MODULES]
+    ana_modules = get_steps_without_pdbs(run_dir, all_steps)
+    log.info(f"Modules not to be analysed: {', '.join(ana_modules)}")
+    sel_step = [st for st in all_steps if st not in ana_modules]
     # check if there are steps to traceback
     if len(sel_step) == 0:
         log.info("No steps to trace back. Exiting.")

src/haddock/libs/libworkflow.py

@@ -67,8 +67,11 @@ def postprocess(self) -> None:
                 capri_steps.append(step.order)  # type: ignore
         # call cli_analyse (no need for capri_dicts, it's all precalculated)
         cli_analyse("./", capri_steps, top_cluster=10, format=None, scale=None)
-        # call cli_traceback
-        cli_traceback("./")
+        # call cli_traceback. If it fails, it's not a big deal
+        try:
+            cli_traceback("./")
+        except Exception as e:
+            log.warning(f"Error running traceback: {e}")
 
 
 class Workflow:

tests/test_cli_traceback.py

@@ -3,11 +3,12 @@
 import os
 import shutil
 from pathlib import Path
+import tempfile
 
 import pandas as pd
 import pytest
 
-from haddock.clis.cli_traceback import main
+from haddock.clis.cli_traceback import main, get_steps_without_pdbs
 
 from . import golden_data
 
@@ -39,6 +40,12 @@ def test_main(rigid_json, flexref_json):
     shutil.copy(rigid_json, os.path.join(step_dirs[0], "io.json"))
     shutil.copy(flexref_json, os.path.join(step_dirs[1], "io.json"))
 
+    # create fake, empty pdb files
+    for i in range(1, 5):
+        open(os.path.join(step_dirs[0], f"rigidbody_{i}.pdb"), "w").close()
+    for i in range(1, 3):
+        open(os.path.join(step_dirs[1], f"flexref_{i}.pdb"), "w").close()
+
     # run haddock3-traceback
     main(run_dir)
 
@@ -82,3 +89,33 @@ def test_analysis():
 
     # check traceback folder does not exist
     assert not os.path.isdir(os.path.join(run_dir, "traceback"))
+
+
+def test_get_steps_without_pdbs():
+    """Test get_steps_without_pdbs."""
+    with tempfile.TemporaryDirectory(dir=".") as tmpdir:
+        # build fake run_dir
+        run_dir = Path(tmpdir)
+        steps = ["0_topoaa", "1_rigidbody", "2_caprieval"]
+
+        for st in steps:
+            os.mkdir(Path(run_dir, st))
+
+        # create fake, empty pdb files
+        fake_pdb_path = Path(run_dir, steps[1] , f"rigidbody_1.pdb")
+        open(fake_pdb_path, "w").close()
+
+        # get steps without pdbs
+        obs_steps = get_steps_without_pdbs(run_dir, steps)
+
+        # check steps without pdbs
+        exp_steps = ["0_topoaa", "2_caprieval"]
+        assert obs_steps == exp_steps
+
+        # now we remove the fake pdb file
+        os.remove(fake_pdb_path)
+
+        # get steps without pdbs
+        obs_steps = get_steps_without_pdbs(run_dir, steps)
+        exp_steps = ["0_topoaa", "1_rigidbody", "2_caprieval"]
+        assert obs_steps == exp_steps