From 0b0fd48c4ef36971e94a1139643d406234f84ea4 Mon Sep 17 00:00:00 2001
From: TJ Lane <thomas.joseph.lane@gmail.com>
Date: Sat, 12 Jun 2021 17:22:10 +0200
Subject: [PATCH] robustness improvements
---
xia2pipe/dbdaemon.py | 2 +-
xia2pipe/dmpldaemon.py | 4 ++--
xia2pipe/projbase.py | 54 ++++++++++++++++++++++++++++++------------
3 files changed, 42 insertions(+), 18 deletions(-)
diff --git a/xia2pipe/dbdaemon.py b/xia2pipe/dbdaemon.py
index f289337..8c04ae1 100644
--- a/xia2pipe/dbdaemon.py
+++ b/xia2pipe/dbdaemon.py
@@ -12,7 +12,7 @@ class DBDaemon(ProjectBase):
def in_db(self, metadata, run, table):
- cid = self.metadata_to_id(metadata, run)
+ cid = self.metadata_to_id(metadata, run) # crystal ID
if table == 'Data_Reduction':
try:
diff --git a/xia2pipe/dmpldaemon.py b/xia2pipe/dmpldaemon.py
index 0994d21..f24b1e8 100644
--- a/xia2pipe/dmpldaemon.py
+++ b/xia2pipe/dmpldaemon.py
@@ -166,7 +166,7 @@ class DimplingDaemon(ProjectBase):
#SBATCH --ntasks=1
#SBATCH --cpus-per-task={nproc}
#SBATCH --mem=6GB
-#SBATCH --time=5:00:00
+#SBATCH --time=10:00:00
#SBATCH --job-name {name}-dmpl_{metadata}-{run}
#SBATCH --output {outdir}/{name}-dmpl_{metadata}-{run}.out
#SBATCH --error {outdir}/{name}-dmpl_{metadata}-{run}.err
@@ -252,7 +252,7 @@ if __name__ == '__main__':
#('l11p17_11', 1),
]
- dd = DimplingDaemon.load_config('../configs/d1p7_v2.yaml')
+ dd = DimplingDaemon.load_config('../configs/mpro_v2/DIALS_v2.yaml')
for md, run in jobs:
print(md, run)
diff --git a/xia2pipe/projbase.py b/xia2pipe/projbase.py
index dd09d51..c51437f 100644
--- a/xia2pipe/projbase.py
+++ b/xia2pipe/projbase.py
@@ -78,22 +78,41 @@ def _phenix_model_b(pdb_path):
def _get_pdb_chosen_by_dimple(dimple_log_path):
+ """
+ dimple is used to choose the closest matching PDB
- pdb_path = None
-
- with open(dimple_log_path, 'r') as f:
-
- watching = False
- for line in f.readlines():
+ this function returns just the PDB file name, not a full path
+ """
- if line.startswith('# PDBs in order of similarity (using the first one):'):
- watching = True
- if line.startswith('#') and watching:
- g = re.search('(\S+\.pdb)', line)
- if g:
- pdb_path = g.groups()[0]
- break
+ # TJL 06 Jan 2021
+ # I think this works and is much better
+ # the old version (commented below) breaks if you have only one PDB input
+
+ config = configparser.ConfigParser()
+ config.read(dimple_log_path)
+
+ pdbs = ast.literal_eval( config['workflow']['pdb_files'] )
+ pdb_path = pdbs[0] # they are in order
+
+ #pdb_path = None
+ #
+ #with open(dimple_log_path, 'r') as f:
+ # watching = False
+ # for line in f.readlines():
+ #
+ # if line.startswith('# PDBs in order of similarity (using the first one):'):
+ # watching = True
+ #
+ # if line.startswith('#') and watching:
+ # g = re.search('(\S+\.pdb)', line)
+ # if g:
+ # pdb_path = g.groups()[0]
+ # break
+ #
+ #if pdb_path is None:
+ # raise IOError('could not find dimple-chosen reference PDB'
+ # ' in: {}'.format(dimple_log_path))
return pdb_path
@@ -139,6 +158,11 @@ class ProjectBase:
self.xia2_config = xia2_config
self.refinement_config = refinement_config
+ # this is for backward compatability and is not desirable
+ # provides a default method name
+ if 'method_name' not in self.refinement_config.keys():
+ self.refinement_config['method_name'] = 'dmpl2'
+
return
@@ -221,7 +245,7 @@ class ProjectBase:
qid = self.db.select('refinement_id',
'{}.Refinement'.format(self._analysis_db),
{'data_reduction_id' : data_reduction_id,
- 'method' : 'dmpl2'})
+ 'method' : self.refinement_config['method_name']})
return get_single(qid, crystal_id, run, 'refinement_id')
@@ -603,7 +627,7 @@ class ProjectBase:
'initial_pdb_path': initial_pdb_path,
'final_pdb_path': pdb_path,
'refinement_mtz_path': mtz_path,
- 'method': 'dmpl2',
+ 'method': self.refinement_config['method_name'],
'resolution_cut': self.get_refinement_res(metadata, run),
'rfree': log_results[1],
'rwork': log_results[0],
--
GitLab