From 15d38abf46fc3afe2a525e3d4ed3acb49d491a94 Mon Sep 17 00:00:00 2001
From: iquasere <maildosequeira@gmail.com>
Date: Tue, 19 Dec 2023 11:42:24 +0000
Subject: [PATCH] ID mapping columns now must be properly inputted If some
 column is not in return fields, UPIMAPI will exit with a code different from
 0 Added parameters for showing fields

---
 README.md  |  8 ++++++--
 upimapi.py | 21 ++++++++++++++++++++-
 2 files changed, 26 insertions(+), 3 deletions(-)

diff --git a/README.md b/README.md
index 73ae5b5..cd83ae8 100644
--- a/README.md
+++ b/README.md
@@ -68,13 +68,17 @@ To input a custom database into UPIMAPI, specify it as ```--database path/to/dat
 
 ### Columns of information from UniProt
 
-UniProt provides information for many different fields of information and cross-references. For the user's convenience, a default selection is provided: ```Entry```, ```Entry name```, ```Gene names```, ```Protein names```, ```EC number```, ```Function[CC]```, ```Pathway```, ```Keywords```, ```Protein existence```, ```Gene ontology (GO)```, ```Protein families```, ```Taxonomic lineage```, ```Organism```, ```Organism ID```, ```BioCyc```, ```BRENDA```, ```CDD```, ```eggNOG```, ```Ensembl```, ```InterPro```, ```KEGG```, ```Pfam```, ```Reactome```, ```RefSeq``` and ```UniPathway```
+UniProt provides information for many different fields of information and cross-references. For the user's convenience, a default selection is provided: ```Entry```, ```Entry name```, ```Gene Names```, ```Protein names```, ```EC number```, ```Function[CC]```, ```Pathway```, ```Keywords```, ```Protein existence```, ```Gene Ontology (GO)```, ```Protein families```, ```Taxonomic lineage```, ```Organism```, ```Organism ID```, ```BioCyc```, ```BRENDA```, ```CDD```, ```eggNOG```, ```Ensembl```, ```InterPro```, ```KEGG```, ```Pfam```, ```Reactome```, ```RefSeq``` and ```UniPathway```.
 
 If another selection of columns/databases is desired, it can be specified, for example, as 
 ```
 --columns "Coiled coil&Compositional bias"
 ```
-where ```--columns``` takes as input the names of the fields of information required. The complete list of fields available can be consulted at [UniProtKB return fields](https://www.uniprot.org/help/return_fields).
+where ```--columns``` takes as input the names of the fields of information required. Valid values for the columns can be consulted at [UniProtKB return fields](https://www.uniprot.org/help/return_fields).
+
+#### Sometimes the return fields are not properly updated
+
+If the columns were correctly inputted according to the [return fields page](https://www.uniprot.org/help/return_fields) and UPIMAPI is still complaining about "\[COL] is not a valid column name for ID mapping", it may be that values at return fields are not properly updated. If that happens, running `upimapi --show-available-columns` will present the user with the current valid fields.
 
 #### UPIMAPI offers a few additional columns for taxonomic information
 
diff --git a/upimapi.py b/upimapi.py
index c8ced33..cc36f7a 100644
--- a/upimapi.py
+++ b/upimapi.py
@@ -28,7 +28,7 @@
 from functools import partial
 import re
 
-__version__ = '1.12.3'
+__version__ = '1.13.0'
 
 
 def load_api_info():
@@ -165,10 +165,29 @@ def get_arguments():
     diamond_args.add_argument(
         '--diamond-mode', help="Mode to run DIAMOND with [fast]", default='fast',
         choices=['fast', 'mid_sensitive', 'sensitive', 'more_sensitive', 'very_sensitive', 'ultra_sensitive'])
+
+    special_functions = parser.add_argument_group('Special functions')
+    special_functions.add_argument(
+        "--show-available-fields", action="store_true", default=False,
+        help="Outputs the fields available from the API.")
+
     args = parser.parse_args()
+    if args.show_available_fields:
+        sys.exit('\n'.join(columns_dict.keys()))
+
     args.output = args.output.rstrip('/')
     args.resources_directory = args.resources_directory.rstrip('/')
     args.columns = args.columns.split('&') if args.columns else None
+
+    columns_fine = True
+    for col in args.columns:
+        if col not in columns_dict.keys():
+            print(
+                f'ERR: [{col}] is not a valid column name for ID mapping. For more information, check '
+                f'https://github.com/iquasere/UPIMAPI/tree/master#sometimes-the-return-fields-are-not-properly-updated')
+            columns_fine = False
+    if not columns_fine:
+        sys.exit(1)
     if args.taxids is not None:
         args.taxids = args.taxids.split(',')
     return args