Added BOF to cmpb pipeline #8

draeger-lab · May 24, 2024 · 37b400c · 37b400c
1 parent 73666da
commit 37b400c
Show file tree

Hide file tree

Showing 4 changed files with 119 additions and 20 deletions.
diff --git a/dev/CB_test_modules.ipynb b/dev/CB_test_modules.ipynb
@@ -2760,6 +2760,89 @@
    "source": [
     "specimen.hqtb."
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "<bound method SBase.getNotesString of <Model Kp_std \"Genome-scale metabolic model of K. pneumoniae HS11286\">>"
+      ]
+     },
+     "execution_count": 2,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "from refinegems.utility.io import load_model\n",
+    "\n",
+    "p = '/Users/brune/Documents/11_Test_Data/test_SPECIMEN/thesis/Kp_std/02_generate_draft_model/Kp_std_draft.xml'\n",
+    "model = load_model(p,'libsbml')\n",
+    "\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 26,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model.getModelHistory()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 24,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "'<notes>\\n  <html xmlns=\"http://www.w3.org/1999/xhtml\">\\n    <p>Description: This model was built with CarveMe version 1.5.1</p>\\n  </html>\\n</notes>'"
+      ]
+     },
+     "execution_count": 24,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "x = model.getNotesString()\n",
+    "x\n",
+    "# 'CarveMe' in x"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 13,
+   "metadata": {},
+   "outputs": [
+    {
+     "data": {
+      "text/plain": [
+       "{'Description': 'This model was built with CarveMe version 1.5.1'}"
+      ]
+     },
+     "execution_count": 13,
+     "metadata": {},
+     "output_type": "execute_result"
+    }
+   ],
+   "source": [
+    "mod2 = load_model(p, 'cobra')\n",
+    "mod2.notes"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": []
   }
  ],
  "metadata": {

diff --git a/src/specimen/cmpb/workflow.py b/src/specimen/cmpb/workflow.py
@@ -6,8 +6,6 @@
 # requirements
 ################################################################################
 
-import os
-import cobra
 import logging
 import pandas as pd
 from datetime import date
@@ -16,11 +14,12 @@
 import warnings
 import yaml
 
+from cobra import Reaction
 from libsbml import readSBML
 
 from refinegems.curation.pathways import kegg_pathway_analysis
 from refinegems.classes.reports import ModelInfoReport
-from refinegems.curation.biomass import test_biomass_presence
+from refinegems.curation.biomass import test_biomass_presence, check_normalise_biomass
 from refinegems.analysis import growth
 from refinegems.utility.connections import run_memote, perform_mcc, adjust_BOF
 from refinegems.curation.curate import resolve_duplicates
@@ -263,20 +262,39 @@ def between_analysis(model, cfg, step):
 
     # BOF
     # ---
-    # @TODO
     # BOFdat - optional
+    current_model = load_model(current_modelpath,'cobra')
     if config['BOF']['run_bofdat']:
-        new_bof = adjust_BOF(config['BOF']['bofdat_params']['full_genome_sequence'], 
-                            current_modelpath,
-                            current_model, 
-                            dna_weight_fraction = config['BOF']['bofdat_params']['dna_weight_fraction'], 
-                            weight_frac = config['BOF']['bofdat_params']['weight_fraction']) 
-        # @TODO 
-        # save ?
-    # check and normalise
-    elif config['BOF']['normalise']:
-        pass
-        #@TODO
+        check_bof = test_biomass_presence(current_model)
+        if check_bof:
+            current_model.reactions.get_by_id(check_bof[0]).reaction = adjust_BOF(config['BOF']['bofdat_params']['full_genome_sequence'], 
+                                current_modelpath,
+                                current_model, 
+                                dna_weight_fraction = config['BOF']['bofdat_params']['dna_weight_fraction'], 
+                                weight_frac = config['BOF']['bofdat_params']['weight_fraction']) 
+            current_model = check_normalise_biomass(current_model)
+        else: 
+            bof_reac = Reaction('Biomass_BOFdat')
+            bof_reac.name = 'Biomass objective function created by BOFdat'
+            current_model.add_reactions([bof_reac])
+            current_model.reactions.get_by_id(bof_reac).reaction = adjust_BOF(config['BOF']['bofdat_params']['full_genome_sequence'], 
+                                current_modelpath,
+                                current_model, 
+                                dna_weight_fraction = config['BOF']['bofdat_params']['dna_weight_fraction'], 
+                                weight_frac = config['BOF']['bofdat_params']['weight_fraction'])
+            current_model = check_normalise_biomass(current_model)
+
+    # check and normalise 
+    else:
+        current_model = check_normalise_biomass(current_model)
+
+    # save 
+    # save model
+    if config['general']['save_all_models']:
+        write_model_to_file(Path(dir,'cmpb_out','models','after_BOF.xml'))
+        current_modelpath = Path(dir,'cmpb_out','models','after_BOF.xml')
+    else:
+        write_model_to_file(current_modelpath)
 
     # MCC
     # ---

diff --git a/src/specimen/data/config/cmpb_config.yaml b/src/specimen/data/config/cmpb_config.yaml
@@ -86,7 +86,5 @@ BOF:
     full_genome_sequence: USER  # whole genome sequence
     dna_weight_fraction: USER   # DNA weight fraction for the organism
     weight_fraction: USER       # Ezyme/ion weight fractions for the organism
-  # if BOFdat is not run, the biomass objective function
-  # can still - optionally - normalised
-  normalise: True 
-  # @TODO
+
+
diff --git a/src/specimen/hqtb/core/refinement/smoothing.py b/src/specimen/hqtb/core/refinement/smoothing.py
@@ -163,7 +163,7 @@ def run(genome:str,model:str,dir:str,mcc='skip',
             bof_reac = Reaction('Biomass_BOFdat')
             bof_reac.name = 'Biomass objective function created by BOFdat'
             model.add_reactions([bof_reac])
-            model.reactions.get_by_id(pos_bofs[0]).reaction = adjust_BOF(genome, temp_model.name, model, dna_weight_frac, ion_weight_frac)
+            model.reactions.get_by_id(bof_reac).reaction = adjust_BOF(genome, temp_model.name, model, dna_weight_frac, ion_weight_frac)
 
             # optimise BOF(s)
             model = check_normalise_biomass(model)