gsrap 0.8.0__tar.gz → 0.8.1__tar.gz

{gsrap-0.8.0 → gsrap-0.8.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: gsrap
-Version: 0.8.0
+Version: 0.8.1
 Summary: 
 License: GNU General Public License v3.0
 Author: Gioele Lazzari

{gsrap-0.8.0 → gsrap-0.8.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "gsrap"
-version = "0.8.0"
+version = "0.8.1"
 description = ""
 authors = ["Gioele Lazzari"]
 license = "GNU General Public License v3.0"

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/.ipynb_checkpoints/__init__-checkpoint.py

@@ -72,7 +72,7 @@ def main():
     parsedb_parser.add_argument("--precursors", action='store_true', help="Verify biosynthesis of biomass precursors and show blocked ones.")
     parsedb_parser.add_argument("--biosynth", action='store_true', help="Check biosynthesis of all metabolites and detect dead-ends.")
     parsedb_parser.add_argument("-e", "--eggnog", nargs='+', metavar='', type=str, default='-', help="Path to the optional eggnog-mapper annotation table(s).")
-    parsedb_parser.add_argument("-k", "--keggorg", metavar='', type=str, default='-', help="A single KEGG Organism code.")
+    parsedb_parser.add_argument("-k", "--keggorg", metavar='', type=str, default='-', help="A single KEGG Organism code. If provided, it takes precedence over --eggnog.")
     parsedb_parser.add_argument("--goodbefore", metavar='', type=str, default='-', help="Syntax is {pure_mid}-{rid1}-{rid2}. From top to bottom, build the universe until reaction {rid1}, transport {rid2} and metabolite {pure_mid} are reached.")
     parsedb_parser.add_argument("--onlyauthor", metavar='', type=str, default='-', help="Build the universe by parsing contents of the specified author ID only. Contents affected by --goodbefore are parsed anyway.")
     parsedb_parser.add_argument("--nofigs", action='store_true', help="Do not generate figures.")
@@ -85,7 +85,7 @@ def main():
     mkmodel_parser.add_argument("-c", "--cores", metavar='', type=int, default=0, help="Number of cores to use (if 0, use all available cores).")
     mkmodel_parser.add_argument("-o", "--outdir", metavar='', type=str, default='./', help="Main output directory (will be created if not existing).")
     mkmodel_parser.add_argument("-e", "--eggnog", nargs='+', metavar='', type=str, default='-', help="Path to the eggnog-mapper annotation table(s).")
-    mkmodel_parser.add_argument("-k", "--keggorg", metavar='', type=str, default='-', help="A single KEGG Organism code.")
+    mkmodel_parser.add_argument("-k", "--keggorg", metavar='', type=str, default='-', help="A single KEGG Organism code. If provided, it takes precedence over --eggnog.")
     mkmodel_parser.add_argument("-u", "--universe", metavar='', type=str, default='-', help="Path to the universe model (SBML format).")
     mkmodel_parser.add_argument("-i", "--force_inclusion", metavar='', type=str, default='-', help="Force the inclusion of the specified reactions (comma-separated IDs).")
     mkmodel_parser.add_argument("-f", "--gap_fill", metavar='', type=str, default='-', help="Media to use during gap-filling (comma-separated IDs); if not provided, gap-filling will be skipped.")

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/__init__.py

@@ -72,7 +72,7 @@ def main():
     parsedb_parser.add_argument("--precursors", action='store_true', help="Verify biosynthesis of biomass precursors and show blocked ones.")
     parsedb_parser.add_argument("--biosynth", action='store_true', help="Check biosynthesis of all metabolites and detect dead-ends.")
     parsedb_parser.add_argument("-e", "--eggnog", nargs='+', metavar='', type=str, default='-', help="Path to the optional eggnog-mapper annotation table(s).")
-    parsedb_parser.add_argument("-k", "--keggorg", metavar='', type=str, default='-', help="A single KEGG Organism code.")
+    parsedb_parser.add_argument("-k", "--keggorg", metavar='', type=str, default='-', help="A single KEGG Organism code. If provided, it takes precedence over --eggnog.")
     parsedb_parser.add_argument("--goodbefore", metavar='', type=str, default='-', help="Syntax is {pure_mid}-{rid1}-{rid2}. From top to bottom, build the universe until reaction {rid1}, transport {rid2} and metabolite {pure_mid} are reached.")
     parsedb_parser.add_argument("--onlyauthor", metavar='', type=str, default='-', help="Build the universe by parsing contents of the specified author ID only. Contents affected by --goodbefore are parsed anyway.")
     parsedb_parser.add_argument("--nofigs", action='store_true', help="Do not generate figures.")
@@ -85,7 +85,7 @@ def main():
     mkmodel_parser.add_argument("-c", "--cores", metavar='', type=int, default=0, help="Number of cores to use (if 0, use all available cores).")
     mkmodel_parser.add_argument("-o", "--outdir", metavar='', type=str, default='./', help="Main output directory (will be created if not existing).")
     mkmodel_parser.add_argument("-e", "--eggnog", nargs='+', metavar='', type=str, default='-', help="Path to the eggnog-mapper annotation table(s).")
-    mkmodel_parser.add_argument("-k", "--keggorg", metavar='', type=str, default='-', help="A single KEGG Organism code.")
+    mkmodel_parser.add_argument("-k", "--keggorg", metavar='', type=str, default='-', help="A single KEGG Organism code. If provided, it takes precedence over --eggnog.")
     mkmodel_parser.add_argument("-u", "--universe", metavar='', type=str, default='-', help="Path to the universe model (SBML format).")
     mkmodel_parser.add_argument("-i", "--force_inclusion", metavar='', type=str, default='-', help="Force the inclusion of the specified reactions (comma-separated IDs).")
     mkmodel_parser.add_argument("-f", "--gap_fill", metavar='', type=str, default='-', help="Media to use during gap-filling (comma-separated IDs); if not provided, gap-filling will be skipped.")

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/commons/.ipynb_checkpoints/medium-checkpoint.py

@@ -17,10 +17,9 @@ def apply_medium_given_column(logger, model, medium, column, is_reference=False)
     column = column.to_dict()
 
     
-    # add trace elements:
-    column['fe2'] = 'NL'
-    column['mobd'] = 'NL'
-    column['cobalt2'] = 'NL'
+    # add default elements (acqueous media)
+    column['h2o'] = 'NL'
+    column['h'] = '-0.0001'  # pH=7
 
 
     # reset exchanges

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/commons/medium.py

@@ -17,10 +17,9 @@ def apply_medium_given_column(logger, model, medium, column, is_reference=False)
     column = column.to_dict()
 
     
-    # add trace elements:
-    column['fe2'] = 'NL'
-    column['mobd'] = 'NL'
-    column['cobalt2'] = 'NL'
+    # add default elements (acqueous media)
+    column['h2o'] = 'NL'
+    column['h'] = '-0.0001'  # pH=7
 
 
     # reset exchanges

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/mkmodel/.ipynb_checkpoints/mkmodel-checkpoint.py

@@ -44,6 +44,9 @@ from ..commons import download_keggorg
 
 from ..runsims.biosynth import biosynthesis_on_media
 
+from ..parsedb.cycles import verify_egc_all
+
+
 
 
 def create_model_incore(params):
@@ -140,6 +143,9 @@ def create_model_incore(params):
 
     
     ###### CHECKS
+    # check erroneous EGCs
+    verify_egc_all(logger, model, args.outdir)
+    
     # check blocked metabolites / dead-ends
     df_S = biosynthesis_on_media(logger, model, dbexp, args.gap_fill, args.biosynth)
     if type(df_S)==int: return 1

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/mkmodel/mkmodel.py

@@ -44,6 +44,9 @@ from ..commons import download_keggorg
 
 from ..runsims.biosynth import biosynthesis_on_media
 
+from ..parsedb.cycles import verify_egc_all
+
+
 
 
 def create_model_incore(params):
@@ -140,6 +143,9 @@ def create_model_incore(params):
 
     
     ###### CHECKS
+    # check erroneous EGCs
+    verify_egc_all(logger, model, args.outdir)
+    
     # check blocked metabolites / dead-ends
     df_S = biosynthesis_on_media(logger, model, dbexp, args.gap_fill, args.biosynth)
     if type(df_S)==int: return 1

gsrap-0.8.1/src/gsrap/parsedb/.ipynb_checkpoints/cycles-checkpoint.py

@@ -0,0 +1,128 @@
+import warnings 
+import os
+import logging
+
+
+import cobra
+import gempipe
+
+
+from ..commons import fba_no_warnings
+from ..commons import get_optthr
+
+
+
+def verify_egc(logger, model, mid, outdir): 
+
+    
+    # changes as not permament: 
+    found_egc = False
+    with model: 
+        
+        # close (0; 0) all the exchange reactions: 
+        gempipe.close_boundaries(model)
+        
+                
+        # create a dissipation reaction: 
+        dissip = cobra.Reaction(f'__dissip__{mid}')
+        model.add_reactions([dissip])
+        dissip = model.reactions.get_by_id(f'__dissip__{mid}')
+        
+        
+        # define the dissipation reaction:
+        modeled_mids = [m.id for m in model.metabolites]
+        if   mid == 'atp':
+            dissip_string = 'atp_c + h2o_c --> adp_c + pi_c + h_c'
+        elif mid == 'ctp':
+            dissip_string = 'ctp_c + h2o_c --> cdp_c + pi_c + h_c'
+        elif mid == 'gtp':
+            dissip_string = 'gtp_c + h2o_c --> gdp_c + pi_c + h_c'
+        elif mid == 'utp':
+            dissip_string = 'utp_c + h2o_c --> udp_c + pi_c + h_c'
+        elif mid == 'itp':
+            dissip_string = 'itp_c + h2o_c --> idp_c + pi_c + h_c'
+        elif mid == 'nadh':
+            dissip_string = 'nadh_c --> nad_c + h_c'
+        elif mid == 'nadph':
+            dissip_string = 'nadph_c --> nadp_c + h_c'
+        elif mid == 'fadh2':
+            dissip_string = 'fadh2_c --> fad_c + 2.0 h_c'
+        elif mid == 'accoa':
+            dissip_string = 'accoa_c + h2o_c --> ac_c + coa_c + h_c'
+        elif mid == 'glu__L':
+            dissip_string = 'glu__L_c + h2o_c --> akg_c + nh4_c + 2.0 h_c'
+        elif mid == 'q8h2':
+            dissip_string = 'q8h2_c --> q8_c + 2.0 h_c'
+        dissip.build_reaction_from_string(dissip_string)
+        
+        
+        # set the objective and optimize: 
+        model.objective = f'__dissip__{mid}'
+        res, obj_value, status = fba_no_warnings(model)
+        
+        
+        # apply the threshold:
+        obj_value = res.objective_value
+        status = res.status
+        if status == 'optimal' and obj_value >= get_optthr():
+            found_egc = True
+                
+
+            # get suspect !=0 fluxes 
+            fluxes = res.fluxes
+            # get interesting fluxes (get_optthr() tries to take into account the approximation in glpk and cplex solvers)
+            fluxes_interesting = fluxes[(fluxes > get_optthr()) | (fluxes < -get_optthr())]
+            
+
+            # create a model for escher, remove Rs not beloning to the cycle
+            model_copy = model.copy()
+            all_rids = [r.id for r in model_copy.reactions]
+            to_delete = set(all_rids) - set(fluxes_interesting.index)
+
+
+            # trick to avoid the WARNING "cobra/core/group.py:147: UserWarning: need to pass in a list" 
+            # triggered when trying to remove reactions that are included in groups. 
+            with warnings.catch_warnings():  # temporarily suppress warnings for this block
+                warnings.simplefilter("ignore")  # ignore all warnings
+                cobra_logger = logging.getLogger("cobra.util.solver")
+                old_level = cobra_logger.level
+                cobra_logger.setLevel(logging.ERROR)   
+                
+                # triggering code
+                model_copy.remove_reactions(to_delete)  # should work also with IDs
+
+                # restore original behaviour: 
+                cobra_logger.setLevel(old_level) 
+                
+                
+            # save JSON to direct import in Escher:
+            outfile = os.path.join(outdir, f'EGC_{mid}.json')
+            cobra.io.save_json_model(model_copy, outfile)
+
+
+            # log some messages
+            rid_labels = []
+            for rid, flux in fluxes_interesting.to_dict().items():
+                rid_label = "'" + rid + "'" 
+                # mark reversible reactions composing the cycle:
+                r = model.reactions.get_by_id(rid)
+                if r.lower_bound < 0 and r.upper_bound > 0:
+                    rid_label = rid_label + '(<=>)'
+                rid_labels.append(rid_label)
+            logger.warning(f"Found erroneous EGC (N={len(model_copy.reactions)}) for '{mid}' (f={obj_value}): [{', '.join(rid_labels)}]. EGC saved to '{outfile}' to be inspected with Escher-FBA.")
+            
+            
+    return found_egc
+
+
+
+def verify_egc_all(logger, model, outdir='./', mids_to_check=['atp','ctp','gtp','utp','itp','nadh','nadph','fadh2','accoa','glu__L','q8h2']):
+
+    
+    all_results = []
+    for mid in mids_to_check:
+        all_results.append(verify_egc(logger, model, mid, outdir))
+    if any(all_results)==False:
+        logger.info("Found 0 erroneous energy-generating cycles (EGCs).")
+        
+        

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/parsedb/.ipynb_checkpoints/introduce-checkpoint.py

@@ -287,15 +287,7 @@ def introduce_transporters(logger, db, model, idcollection_dict, kegg_reaction_t
         r = model.reactions.get_by_id(f'EX_{mid_e}')
         r.name = f"Exchange for {model.metabolites.get_by_id(mid_e).name}"
         r.build_reaction_from_string(f'{mid_e} --> ')
-        if mid_e in [
-            # basics:
-            'glc__D_e', 'nh4_e', 'pi_e', 'so4_e', 'h2o_e', 'h_e', 'o2_e', 'co2_e',
-            # metals: 
-            'cu2_e', 'mobd_e', 'fe2_e', 'cobalt2_e',
-        ]:
-            r.bounds = (-1000, 1000)
-        else:
-            r.bounds = (0, 1000)
+        r.bounds = (0, 1000)
             
         # add SBO annotation
         r.annotation['sbo'] = ['SBO:0000627']  # exchange reaction    

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/parsedb/.ipynb_checkpoints/manual-checkpoint.py

@@ -19,6 +19,33 @@ def get_rids_with_mancheck_gpr():
     return rids_mancheck_gpr
 
 
+def get_rids_with_mancheck_balancing():
+    rids_mancheck_bal = [  # same reactions involving ATP can be reversible
+        
+        # SECTION "reversible both in KEGG and MetaCyc"
+        'PGK', 'SUCOAS', 'ADK1', 'GK1', 'NNATr', 'CYTK1', 'ACKr',
+        'DGK1', 'PPAKr', 'ATPSr', 'NDPK10',
+        
+        ### SECTION "reversible in KEGG but not in MetaCyc" ###
+        'CYTK2',  # clearly reversible in KEGG but not in MetaCyc (RXN-7913)
+        'DADK',  # clearly reversible in KEGG but not in MetaCyc (DEOXYADENYLATE-KINASE-RXN)
+        'UMPK',  # clearly reversible in KEGG but not in MetaCyc (RXN-12002)
+        'NDPK1',  # clearly reversible in KEGG but not in MetaCyc (GDPKIN-RXN)
+        'NDPK2',  # clearly reversible in KEGG but not in MetaCyc (UDPKIN-RXN)  
+        'NDPK3',  # clearly reversible in KEGG but not in MetaCyc (CDPKIN-RXN)
+        'NDPK4',  # clearly reversible in KEGG but not in MetaCyc (DTDPKIN-RXN)
+        'NDPK5',  # clearly reversible in KEGG but not in MetaCyc (DGDPKIN-RXN)
+        'NDPK6',  # clearly reversible in KEGG but not in MetaCyc (DUDPKIN-RXN)
+        'NDPK7',  # clearly reversible in KEGG but not in MetaCyc (DCDPKIN-RXN)
+        'NDPK8',  # clearly reversible in KEGG but not in MetaCyc (DADPKIN-RXN)
+        'NDPK9',  # clearly reversible in KEGG but not in MetaCyc (RXN-14120) 
+        
+        ### SECTION "missing reversibility info" ###
+        'LPHERA',  
+    ]
+    return rids_mancheck_bal
+
+
 
 def get_manual_sinks():
     

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/parsedb/.ipynb_checkpoints/parsedb-checkpoint.py

@@ -35,6 +35,8 @@ from ..runsims.biosynth import biosynthesis_on_media
 
 from ..mkmodel.polishing import remove_disconnected
 
+from .cycles import verify_egc_all
+
 
 
 
@@ -173,6 +175,9 @@ def main(args, logger):
     
     
     ###### CHECKS 2
+    # check erroneous EGCs
+    verify_egc_all(logger, universe, args.outdir)
+    
     # check growth on minmal media
     df_G = grow_on_media(logger, universe, dbexp, args.media, '-', True)
     if type(df_G)==int: return 1

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/parsedb/.ipynb_checkpoints/repeating-checkpoint.py

@@ -4,6 +4,7 @@ import cobra
 
 from .manual import get_deprecated_kos
 from .manual import get_rids_with_mancheck_gpr
+from .manual import get_rids_with_mancheck_balancing
 
 
 
@@ -138,6 +139,14 @@ def add_reaction(logger, model, rid, row, kr_ids, kegg_reaction_to_others, addty
         return 1    
     
     
+    # check if reversible and using ATP
+    if r.lower_bound < 0 and r.upper_bound > 0:
+        for m in r.metabolites: 
+            if m.id.rsplit('_', 1)[0] == 'atp':
+                if rid not in get_rids_with_mancheck_balancing():
+                    logger.warning(f"Reaction '{rid}' involves ATP and is reversible: are you sure?")
+    
+    
     return 0
 
 

gsrap-0.8.1/src/gsrap/parsedb/cycles.py

@@ -0,0 +1,128 @@
+import warnings 
+import os
+import logging
+
+
+import cobra
+import gempipe
+
+
+from ..commons import fba_no_warnings
+from ..commons import get_optthr
+
+
+
+def verify_egc(logger, model, mid, outdir): 
+
+    
+    # changes as not permament: 
+    found_egc = False
+    with model: 
+        
+        # close (0; 0) all the exchange reactions: 
+        gempipe.close_boundaries(model)
+        
+                
+        # create a dissipation reaction: 
+        dissip = cobra.Reaction(f'__dissip__{mid}')
+        model.add_reactions([dissip])
+        dissip = model.reactions.get_by_id(f'__dissip__{mid}')
+        
+        
+        # define the dissipation reaction:
+        modeled_mids = [m.id for m in model.metabolites]
+        if   mid == 'atp':
+            dissip_string = 'atp_c + h2o_c --> adp_c + pi_c + h_c'
+        elif mid == 'ctp':
+            dissip_string = 'ctp_c + h2o_c --> cdp_c + pi_c + h_c'
+        elif mid == 'gtp':
+            dissip_string = 'gtp_c + h2o_c --> gdp_c + pi_c + h_c'
+        elif mid == 'utp':
+            dissip_string = 'utp_c + h2o_c --> udp_c + pi_c + h_c'
+        elif mid == 'itp':
+            dissip_string = 'itp_c + h2o_c --> idp_c + pi_c + h_c'
+        elif mid == 'nadh':
+            dissip_string = 'nadh_c --> nad_c + h_c'
+        elif mid == 'nadph':
+            dissip_string = 'nadph_c --> nadp_c + h_c'
+        elif mid == 'fadh2':
+            dissip_string = 'fadh2_c --> fad_c + 2.0 h_c'
+        elif mid == 'accoa':
+            dissip_string = 'accoa_c + h2o_c --> ac_c + coa_c + h_c'
+        elif mid == 'glu__L':
+            dissip_string = 'glu__L_c + h2o_c --> akg_c + nh4_c + 2.0 h_c'
+        elif mid == 'q8h2':
+            dissip_string = 'q8h2_c --> q8_c + 2.0 h_c'
+        dissip.build_reaction_from_string(dissip_string)
+        
+        
+        # set the objective and optimize: 
+        model.objective = f'__dissip__{mid}'
+        res, obj_value, status = fba_no_warnings(model)
+        
+        
+        # apply the threshold:
+        obj_value = res.objective_value
+        status = res.status
+        if status == 'optimal' and obj_value >= get_optthr():
+            found_egc = True
+                
+
+            # get suspect !=0 fluxes 
+            fluxes = res.fluxes
+            # get interesting fluxes (get_optthr() tries to take into account the approximation in glpk and cplex solvers)
+            fluxes_interesting = fluxes[(fluxes > get_optthr()) | (fluxes < -get_optthr())]
+            
+
+            # create a model for escher, remove Rs not beloning to the cycle
+            model_copy = model.copy()
+            all_rids = [r.id for r in model_copy.reactions]
+            to_delete = set(all_rids) - set(fluxes_interesting.index)
+
+
+            # trick to avoid the WARNING "cobra/core/group.py:147: UserWarning: need to pass in a list" 
+            # triggered when trying to remove reactions that are included in groups. 
+            with warnings.catch_warnings():  # temporarily suppress warnings for this block
+                warnings.simplefilter("ignore")  # ignore all warnings
+                cobra_logger = logging.getLogger("cobra.util.solver")
+                old_level = cobra_logger.level
+                cobra_logger.setLevel(logging.ERROR)   
+                
+                # triggering code
+                model_copy.remove_reactions(to_delete)  # should work also with IDs
+
+                # restore original behaviour: 
+                cobra_logger.setLevel(old_level) 
+                
+                
+            # save JSON to direct import in Escher:
+            outfile = os.path.join(outdir, f'EGC_{mid}.json')
+            cobra.io.save_json_model(model_copy, outfile)
+
+
+            # log some messages
+            rid_labels = []
+            for rid, flux in fluxes_interesting.to_dict().items():
+                rid_label = "'" + rid + "'" 
+                # mark reversible reactions composing the cycle:
+                r = model.reactions.get_by_id(rid)
+                if r.lower_bound < 0 and r.upper_bound > 0:
+                    rid_label = rid_label + '(<=>)'
+                rid_labels.append(rid_label)
+            logger.warning(f"Found erroneous EGC (N={len(model_copy.reactions)}) for '{mid}' (f={obj_value}): [{', '.join(rid_labels)}]. EGC saved to '{outfile}' to be inspected with Escher-FBA.")
+            
+            
+    return found_egc
+
+
+
+def verify_egc_all(logger, model, outdir='./', mids_to_check=['atp','ctp','gtp','utp','itp','nadh','nadph','fadh2','accoa','glu__L','q8h2']):
+
+    
+    all_results = []
+    for mid in mids_to_check:
+        all_results.append(verify_egc(logger, model, mid, outdir))
+    if any(all_results)==False:
+        logger.info("Found 0 erroneous energy-generating cycles (EGCs).")
+        
+        

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/parsedb/introduce.py

@@ -287,15 +287,7 @@ def introduce_transporters(logger, db, model, idcollection_dict, kegg_reaction_t
         r = model.reactions.get_by_id(f'EX_{mid_e}')
         r.name = f"Exchange for {model.metabolites.get_by_id(mid_e).name}"
         r.build_reaction_from_string(f'{mid_e} --> ')
-        if mid_e in [
-            # basics:
-            'glc__D_e', 'nh4_e', 'pi_e', 'so4_e', 'h2o_e', 'h_e', 'o2_e', 'co2_e',
-            # metals: 
-            'cu2_e', 'mobd_e', 'fe2_e', 'cobalt2_e',
-        ]:
-            r.bounds = (-1000, 1000)
-        else:
-            r.bounds = (0, 1000)
+        r.bounds = (0, 1000)
             
         # add SBO annotation
         r.annotation['sbo'] = ['SBO:0000627']  # exchange reaction    

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/parsedb/manual.py

@@ -19,6 +19,33 @@ def get_rids_with_mancheck_gpr():
     return rids_mancheck_gpr
 
 
+def get_rids_with_mancheck_balancing():
+    rids_mancheck_bal = [  # same reactions involving ATP can be reversible
+        
+        # SECTION "reversible both in KEGG and MetaCyc"
+        'PGK', 'SUCOAS', 'ADK1', 'GK1', 'NNATr', 'CYTK1', 'ACKr',
+        'DGK1', 'PPAKr', 'ATPSr', 'NDPK10',
+        
+        ### SECTION "reversible in KEGG but not in MetaCyc" ###
+        'CYTK2',  # clearly reversible in KEGG but not in MetaCyc (RXN-7913)
+        'DADK',  # clearly reversible in KEGG but not in MetaCyc (DEOXYADENYLATE-KINASE-RXN)
+        'UMPK',  # clearly reversible in KEGG but not in MetaCyc (RXN-12002)
+        'NDPK1',  # clearly reversible in KEGG but not in MetaCyc (GDPKIN-RXN)
+        'NDPK2',  # clearly reversible in KEGG but not in MetaCyc (UDPKIN-RXN)  
+        'NDPK3',  # clearly reversible in KEGG but not in MetaCyc (CDPKIN-RXN)
+        'NDPK4',  # clearly reversible in KEGG but not in MetaCyc (DTDPKIN-RXN)
+        'NDPK5',  # clearly reversible in KEGG but not in MetaCyc (DGDPKIN-RXN)
+        'NDPK6',  # clearly reversible in KEGG but not in MetaCyc (DUDPKIN-RXN)
+        'NDPK7',  # clearly reversible in KEGG but not in MetaCyc (DCDPKIN-RXN)
+        'NDPK8',  # clearly reversible in KEGG but not in MetaCyc (DADPKIN-RXN)
+        'NDPK9',  # clearly reversible in KEGG but not in MetaCyc (RXN-14120) 
+        
+        ### SECTION "missing reversibility info" ###
+        'LPHERA',  
+    ]
+    return rids_mancheck_bal
+
+
 
 def get_manual_sinks():
     

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/parsedb/parsedb.py

@@ -35,6 +35,8 @@ from ..runsims.biosynth import biosynthesis_on_media
 
 from ..mkmodel.polishing import remove_disconnected
 
+from .cycles import verify_egc_all
+
 
 
 
@@ -173,6 +175,9 @@ def main(args, logger):
     
     
     ###### CHECKS 2
+    # check erroneous EGCs
+    verify_egc_all(logger, universe, args.outdir)
+    
     # check growth on minmal media
     df_G = grow_on_media(logger, universe, dbexp, args.media, '-', True)
     if type(df_G)==int: return 1

{gsrap-0.8.0 → gsrap-0.8.1}/src/gsrap/parsedb/repeating.py

@@ -4,6 +4,7 @@ import cobra
 
 from .manual import get_deprecated_kos
 from .manual import get_rids_with_mancheck_gpr
+from .manual import get_rids_with_mancheck_balancing
 
 
 
@@ -138,6 +139,14 @@ def add_reaction(logger, model, rid, row, kr_ids, kegg_reaction_to_others, addty
         return 1    
     
     
+    # check if reversible and using ATP
+    if r.lower_bound < 0 and r.upper_bound > 0:
+        for m in r.metabolites: 
+            if m.id.rsplit('_', 1)[0] == 'atp':
+                if rid not in get_rids_with_mancheck_balancing():
+                    logger.warning(f"Reaction '{rid}' involves ATP and is reversible: are you sure?")
+    
+    
     return 0