add exluded to outpout, remove manual deprecation

Author: tdudgeon

docs/source/USER-GUIDE.md

@@ -12,6 +12,7 @@ _XChemAlign_ is a small suite of tools for preparing PDB models for loading into
 There are a few steps involved.
 1. [**Enable**](#enabling-the-xchemalign-environment) the XChemAlign environment
 2. [**Declare**](#declaring-things) a few things about your data in two structured files in `yaml`[^1]
+2. [**Declare**](#declaring-things) a few things about your data in two structured files in `yaml`[^1]
 3. [**Collate**](#collating-files) your files in a new (specific) directory structure
 4. [**Align**](#aligning-everything) all binding sites to common origins
 5. [**Release**](#upload-to-fragalysis) the data to Fragalysis
@@ -347,19 +348,8 @@ one for each ligand).
 
 #### Overriding status
 
-There is a mechanism to manually override the status of any crystal. Typically you would use this to deprecate a crystal
-that had previously been included, but you now want to exclude from analysis in Fragalysis. To do this add a section like
-this at the top level of your `config.yaml` file:
-
-```yaml
-overrides:
-  crystals:
-    A71EV2A-x0202:
-      status: deprecated
-      status_reason: bad karma
-```
-
-That example would mark the crystal `A71EV2A-x0202` as being of `deprecated` status.
+There was previously a mechanism to manually override the status of any crystal. This feature has now been removed.
+Instead you should set the status is soakDB to `7 - Analysed & Rejected`.
 
 ### 2.3. The assemblies YAML
 

src/xchemalign/collator.py

@@ -167,6 +167,7 @@ def __init__(self, working_dir, log_file=None, log_level=0, include_git_info=Fal
         self.meta_history = []
         self.all_xtals = None
         self.rejected_xtals = set()
+        self.all_excludes = set()
         self.new_or_updated_xtals = None
         self.compound_codes = {}
         self.compound_smiles = {}
@@ -185,7 +186,10 @@ def __init__(self, working_dir, log_file=None, log_level=0, include_git_info=Fal
             for input in inputs:
                 # Determine which datasets to exclude
                 excluded_datasets = utils.find_property(input, Constants.CONFIG_EXCLUDE)
-                if not excluded_datasets:
+                if excluded_datasets:
+                    for excl in excluded_datasets:
+                        self.all_excludes.add(excl)
+                else:
                     excluded_datasets = []
 
                 input_path = utils.find_path(input, Constants.CONFIG_DIR)
@@ -386,6 +390,9 @@ def validate_data(self):
         if git_info:
             meta[Constants.META_GIT_INFO] = git_info
 
+        if self.all_excludes:
+            meta[Constants.META_EXClUDES] = list(self.all_excludes)
+
         tooltips = {}
         for input in self.inputs:
             code_prefix = input.code_prefix
@@ -1242,21 +1249,6 @@ def _munge_history(self, meta):
                 new_or_updated_xtals[xtal_name] = xtal_data
             all_xtals[xtal_name] = xtal_data
 
-            # look for any user defined deprecations
-            xtal_override = overrides.get(Constants.META_XTALS, {}).get(xtal_name, {})
-            if xtal_override:
-                status_override = xtal_override.get(Constants.META_STATUS)
-                if status_override:
-                    xtal_data[Constants.META_STATUS] = status_override
-                    reason = xtal_override.get(Constants.META_STATUS_REASON)
-                    self.logger.info(
-                        "status for xtal {} is overridden by user to be {}".format(xtal_name, status_override)
-                    )
-                    if reason:
-                        xtal_data[Constants.META_STATUS_REASON] = reason
-                    else:
-                        self.logger.warn("status is overridden, but no reason was given")
-
         self.logger.info("metadata {} has {} items".format(count, total))
         self.logger.info(
             "munging resulted in {} total xtals, {} are new or updated".format(
@@ -1453,6 +1445,7 @@ def main():
                 + str(wd)
             )
         else:  # couldn't even create the Collator object so log file will contain nothing useful
+            print(tb)
             print(
                 "Unexpected fatal error occurred, most likely the configuration is wrong or collator was invoked "
                 + "incorrectly. Please send the command you ran and your current directory to the developers."

src/xchemalign/utils.py

@@ -76,6 +76,7 @@ class Constants:
     META_STATUS_NEW = "new"
     META_STATUS_DEPRECATED = "deprecated"
     META_STATUS_REASON = "status_reason"
+    META_EXClUDES = "excluded_crystals"
     META_XTALS = "crystals"
     META_REFERENCE = "reference"
     META_FILE = "file"