03 Organism Mapping
Jupyter notebook from the ENIGMA Carbon Census 1 project.
NB03 — Organism Mapping (ENIGMA isolates, reaction bridge)¶
Map each compound to the ENIGMA isolate genomes (enigma_genome_depot_enigma, 3,109 genomes) that carry the enzymes for its catabolic reactions. This is the substrate for deliverable (a) — the per-compound isolate utilizer table (NB04).
Why a reaction-level bridge (not pathway-maps). During design we found KEGG pathway-map membership is non-discriminative: broad degradation maps (Degradation of aromatic compounds, Fatty acid degradation, Benzoate degradation) are annotated in ~98% of all 3,109 genomes, so map presence predicts nothing. KEGG reaction (R-number) membership discriminates far better — signature reactions hit 1-130 genomes (<5%) while generic reactions hit ~3,000.
Method. For each compound we take its KEGG reaction set, measure each reaction's genome prevalence across genome_depot, and split reactions into signature (prevalence < 10%, specific steps) vs generic (ubiquitous, uninformative). A genome is called a candidate utilizer only if it carries ≥1 catabolic signature reaction (see filter below) — this is what keeps the call discriminative and directionally correct. Each prediction carries a tier and a completeness fraction.
Catabolic filter (critical). Signature (rare) reactions are frequently specialized biosynthetic steps, and rarity-weighting amplifies them — a genome that can make a compound would otherwise be mis-scored as a predicted degrader. We therefore keep a signature reaction only if it is catabolic for that compound: either (i) the reaction is a member of a KEGG degradation/catabolism-named map (reaction-level membership across 28 such maps), or (ii) it is on a short, hand-curated allowlist of unambiguous catabolic steps that consume the compound but live in mixed (non-degradation-named) maps. Three reactions qualify for the allowlist — R02107 (xanthine→urate, xanthine oxidase), R02612 (phenethylamine→phenylacetaldehyde, deaminating amine oxidase), R07202 (abscisate→8′-hydroxyabscisate, ABA 8′-hydroxylase). Reactions that produce the compound (e.g. R06957 making abscisate, R10597 making 3-hydroxybenzoate from chorismate, R07965/6 making xanthine) are excluded.
Tiers (per RESEARCH_PLAN evidence ladder, now reaction-level).
- T2_pathway — carried catabolic signature reaction is in a KEGG degradation map (catabolic context confirmed; the strongest call short of measured fitness).
- T3_signature — carried catabolic signature reaction is from the curated allowlist (consumes the compound, but no degradation-map context).
Honest coverage ceiling. Only ~67 of 207 distinct reactions across the 54 KEGG-mapped compounds are annotated in any ENIGMA genome; 15 compounds have a signature reaction, but after the catabolic filter only 8 compounds retain a reaction in the consuming (degradative) direction. The remaining 75 compounds get no enzyme-level organism call — the knowledge gap widens at the organism step, which is itself a headline result.
In [1]:
import pandas as pd, numpy as np, json, os, math
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
from pathlib import Path
from dotenv import load_dotenv
DATA = Path('../data'); FIG = Path('../figures')
pd.set_option('display.max_columns', 40); pd.set_option('display.width', 220)
# on-cluster Spark Connect (token from .env; never printed)
load_dotenv('/home/aparkin/BERIL-research-observatory/.env')
_tok = os.environ['KBASE_AUTH_TOKEN']
from pyspark.sql import SparkSession
_url = f'sc://jupyter-aparkin.jupyterhub-prod:15002/;use_ssl=false;x-kbase-token={_tok}'
spark = SparkSession.builder.remote(_url).getOrCreate()
DB = 'enigma_genome_depot_enigma'
print('spark connected:', spark.version)
spark connected: 4.0.1
Load deepened linkage + KEGG reaction sets (from NB02b cache)¶
In [2]:
import re
link = pd.read_csv(DATA / 'compound_linkage_deepened.tsv', sep='\t')
cache = json.loads((DATA / 'deepen_cache.json').read_text())
def _body(url):
e = cache.get(url)
if not e: return None
return e.get('body') if isinstance(e, dict) else e
def rxns_for(cnum):
body = _body(f'https://rest.kegg.jp/link/reaction/{cnum}')
if not body: return []
out = []
for p in body.strip().split('\n'):
f = p.split('\t')
if len(f) == 2 and f[1].startswith('rn:'):
out.append(f[1][3:])
return out
# Reaction-level degradation-map membership: the 28 KEGG maps whose names match
# /degrad|catabol/, expanded to their member reactions (link/reaction/map{XXXXX}).
DEG_RE = re.compile(r'degrad|catabol', re.I)
deg_maps = []
for line in (_body('https://rest.kegg.jp/list/pathway') or '').strip().split('\n'):
f = line.split('\t')
if len(f) == 2 and DEG_RE.search(f[1]):
deg_maps.append(f[0].replace('path:', '').replace('map', ''))
degmap_rxns = set()
for m in deg_maps:
b = _body(f'https://rest.kegg.jp/link/reaction/map{m}')
if not b: continue
for line in b.strip().split('\n'):
f = line.split('\t')
if len(f) == 2 and f[1].startswith('rn:'):
degmap_rxns.add(f[1][3:])
# Curated catabolic allowlist: unambiguous compound-CONSUMING steps in mixed maps.
# R02107 xanthine -> urate (xanthine oxidase; purine catabolism)
# R02612 phenethylamine -> PAA + NH3 (deaminating amine oxidase)
# R07202 abscisate -> 8'-OH-abscisate (ABA 8'-hydroxylase; ABA inactivation)
CATABOLIC_ALLOWLIST = {'R02107', 'R02612', 'R07202'}
kid = link[link['kegg_id'].notna()].copy()
kid['rxn_set'] = kid['kegg_id'].map(rxns_for)
kid['has_degmap'] = kid['kegg_deg_maps'].apply(lambda x: isinstance(x, str) and len(str(x)) > 3)
all_rxn = sorted({r for s in kid['rxn_set'] for r in s})
print('KEGG-mapped compounds:', len(kid))
print('with >=1 reaction :', int((kid['rxn_set'].map(len) > 0).sum()))
print('distinct reactions :', len(all_rxn))
print('degradation maps :', len(deg_maps), '| reactions in degradation maps:', len(degmap_rxns))
KEGG-mapped compounds: 54 with >=1 reaction : 37 distinct reactions : 207 degradation maps : 28 | reactions in degradation maps: 1207
Reaction lookup + genome prevalence¶
Map each R-number to its genome_depot FK id, then count distinct genomes carrying each reaction. Prevalence is the fraction of the 3,109 genomes.
In [3]:
kr = spark.sql(f'SELECT id, kegg_id FROM {DB}.browser_kegg_reaction').toPandas()
r2fk = dict(zip(kr['kegg_id'], kr['id']))
fk2r = {v: k for k, v in r2fk.items()}
fks = [r2fk[r] for r in all_rxn if r in r2fk]
NGEN = spark.sql(f'SELECT COUNT(DISTINCT genome_id) n FROM {DB}.browser_gene').toPandas()['n'][0]
print(f'reactions in genome_depot lookup: {len(fks)}/{len(all_rxn)} total genomes: {NGEN}')
fkstr = ','.join(map(str, fks))
# (reaction_fk, genome_id) carriage for ALL in-genome reactions
carr = spark.sql(f'''
SELECT rr.kegg_reaction_id fk, g.genome_id
FROM {DB}.browser_protein_kegg_reactions rr
JOIN {DB}.browser_gene g ON g.protein_id = rr.protein_id
WHERE rr.kegg_reaction_id IN ({fkstr})
GROUP BY rr.kegg_reaction_id, g.genome_id
''').toPandas()
carr['rxn'] = carr['fk'].map(fk2r)
prev = carr.groupby('rxn')['genome_id'].nunique().rename('n_gen').reset_index()
prev['prev_frac'] = prev['n_gen'] / NGEN
prev_map = dict(zip(prev['rxn'], prev['prev_frac']))
print('carriage rows:', len(carr))
print(prev['prev_frac'].describe().to_string())
reactions in genome_depot lookup: 67/207 total genomes: 3109
carriage rows: 57631 count 67.000000 mean 0.276669 std 0.312125 min 0.000322 25% 0.007076 50% 0.137665 75% 0.566259 max 0.964940
Classify reactions: signature (<10% of genomes), then catabolic filter¶
First split signature (rare, discriminative) vs generic (ubiquitous, dropped). Then keep only catabolic signature reactions — those in a degradation-named map or on the curated allowlist — so biosynthetic 'signatures' can't masquerade as degraders. Each signature reaction below shows its catabolic status (degmap / allowlist / dropped).
In [4]:
SIG = 0.10
prev['kind'] = np.where(prev['prev_frac'] < SIG, 'signature', 'generic')
sig_rxns = set(prev[prev.kind == 'signature']['rxn'])
def cat_source(r):
if r in degmap_rxns: return 'degmap'
if r in CATABOLIC_ALLOWLIST: return 'allowlist'
return ''
prev['cat_source'] = prev['rxn'].map(lambda r: cat_source(r) if r in sig_rxns else '')
cat_rxns = {r for r in sig_rxns if cat_source(r)}
# map each signature reaction to the compound name(s) it serves, for the audit print
rxn2names = {}
for _, r in kid.iterrows():
for x in r['rxn_set']:
if x in sig_rxns:
rxn2names.setdefault(x, set()).add(r['name'])
print(prev['kind'].value_counts().to_string())
print(f'signature reactions: {len(sig_rxns)} -> catabolic (kept): {len(cat_rxns)}'
f' (degmap {sum(1 for r in cat_rxns if r in degmap_rxns)},'
f' allowlist {sum(1 for r in cat_rxns if r in CATABOLIC_ALLOWLIST)})')
print()
print('SIGNATURE REACTIONS (catabolic status; * = dropped as biosynthetic/wrong-direction):')
sg = prev[prev.kind == 'signature'].sort_values('prev_frac')
for _, row in sg.iterrows():
r = row['rxn']; src = row['cat_source'] or '*DROP'
nm = ','.join(sorted(rxn2names.get(r, [])))[:34]
print(f' {r:8s} {row["prev_frac"]*100:6.2f}% {src:9s} {nm}')
kind generic 37 signature 30 signature reactions: 30 -> catabolic (kept): 13 (degmap 10, allowlist 3) SIGNATURE REACTIONS (catabolic status; * = dropped as biosynthetic/wrong-direction): R01274 0.03% *DROP Palmitic Acid R10412 0.03% *DROP Farnesol R09849 0.03% *DROP Farnesol R06957 0.03% *DROP Abscisic acid R07202 0.06% allowlist Abscisic acid R01706 0.06% *DROP Palmitic Acid R10902 0.13% *DROP Tyramine R00697 0.16% *DROP Cinnamic acid R07703 0.26% degmap phthalic acid R11664 0.32% *DROP lauric acid R02255 0.39% *DROP Cinnamic acid R01943 0.39% *DROP caffeic acid R07965 0.58% *DROP xanthine R07966 0.58% *DROP xanthine R01769 0.61% *DROP xanthine R02107 0.61% allowlist xanthine R09227 0.64% degmap phthalic acid R01883 0.77% *DROP GUANIDINEACETIC ACID R02252 0.77% *DROP Cinnamic acid R01294 0.80% degmap 4-hydroxybenzaldehyde R05643 1.83% degmap phthalic acid R05360 2.19% degmap salicylic acid R02612 2.61% allowlist Phenylethylamine R01628 2.96% degmap 3-hydroxybenzoic acid R05148 3.12% degmap TEREPHTHALIC ACID R10597 3.67% *DROP 3-hydroxybenzoic acid R02941 4.21% degmap salicylic acid R00565 4.60% *DROP GUANIDINEACETIC ACID R02675 5.34% degmap 4-hydroxybenzaldehyde R05375 7.24% degmap 3-hydroxybenzoic acid
Per-compound reaction coverage¶
How many compounds even reach the organism step — the honest funnel.
In [5]:
rows = []
for _, r in kid.iterrows():
rs = r['rxn_set']
in_gd = [x for x in rs if x in prev_map]
sig = [x for x in in_gd if x in sig_rxns]
cat = [x for x in sig if x in cat_rxns]
cat_dm = [x for x in cat if x in degmap_rxns]
rows.append(dict(compound_id=r['compound_id'], name=r['name'],
npc_pathway=r['npc_pathway'], kegg_id=r['kegg_id'],
n_rxn=len(rs), n_in_gd=len(in_gd), n_sig=len(sig),
n_cat=len(cat), has_cat_degmap=bool(cat_dm),
cat_rxns=';'.join(cat)))
cov = pd.DataFrame(rows)
N = len(link)
print(f'compounds total : {N}')
print(f' with KEGG id : {len(kid)}')
print(f' with >=1 reaction in genome_depot : {int((cov.n_in_gd > 0).sum())}')
print(f' with >=1 signature reaction : {int((cov.n_sig > 0).sum())}')
print(f' with >=1 CATABOLIC signature (callable) : {int((cov.n_cat > 0).sum())}')
print(f' of which degradation-map (T2) : {int((cov.has_cat_degmap).sum())}')
print(f' allowlist-only (T3) : {int(((cov.n_cat > 0) & ~cov.has_cat_degmap).sum())}')
print()
print(cov[cov.n_cat > 0][['name', 'kegg_id', 'n_rxn', 'n_in_gd', 'n_sig', 'n_cat', 'has_cat_degmap']]
.sort_values(['has_cat_degmap', 'n_cat'], ascending=[False, False]).to_string(index=False))
compounds total : 83
with KEGG id : 54
with >=1 reaction in genome_depot : 21
with >=1 signature reaction : 15
with >=1 CATABOLIC signature (callable) : 8
of which degradation-map (T2) : 5
allowlist-only (T3) : 3
name kegg_id n_rxn n_in_gd n_sig n_cat has_cat_degmap
phthalic acid C01606 7 4 3 3 True
4-hydroxybenzaldehyde C00633 12 4 2 2 True
salicylic acid C00805 17 6 2 2 True
3-hydroxybenzoic acid C00587 9 5 3 2 True
TEREPHTHALIC ACID C06337 3 2 1 1 True
xanthine C00385 17 10 4 1 False
Phenylethylamine C05332 4 3 1 1 False
Abscisic acid C06082 6 2 2 1 False
Genome → strain → taxon resolution¶
Load the small entity tables so each predicted genome carries an isolate name and NCBI taxon (used directly by NB04 deliverable a).
In [6]:
gen = spark.sql(f'SELECT id genome_id, name genome_name, strain_id, taxon_id FROM {DB}.browser_genome').toPandas()
strn = spark.sql(f'SELECT id strain_pk, strain_id strain_code, full_name strain_full FROM {DB}.browser_strain').toPandas()
tax = spark.sql(f'SELECT id taxon_pk, taxonomy_id ncbi_taxid, rank, name taxon_name FROM {DB}.browser_taxon').toPandas()
gmap = gen.merge(strn, left_on='strain_id', right_on='strain_pk', how='left') \
.merge(tax, left_on='taxon_id', right_on='taxon_pk', how='left')
gmap = gmap[['genome_id', 'genome_name', 'strain_full', 'ncbi_taxid', 'rank', 'taxon_name']]
print('genome metadata rows:', len(gmap))
print(gmap.head(5).to_string(index=False))
genome metadata rows: 3110
genome_id genome_name strain_full ncbi_taxid rank taxon_name
1 2APBS1 Rhodanobacter denitrificans 666685 species Rhodanobacter denitrificans
2 acidovorax_3H11 Acidovorax sp. GW101-3H11 1813946 species Acidovorax sp. GW101-3H11
3 Agro Agrobacterium fabrum str. C58 176299 strain Agrobacterium fabrum str. C58
4 ANA3 Shewanella sp. ANA-3 94122 species Shewanella sp. ANA-3
5 azobra Azospirillum brasilense Sp245 1064539 species Azospirillum baldaniorum
Build per-(compound, genome) predictions¶
Emit a row only when the genome carries ≥1 signature reaction of the compound. Score = sum of -log10(prevalence) over carried signature reactions (rarity-weighted). sig_completeness = carried signatures / total signatures for that compound.
In [7]:
# reaction -> compounds (a CATABOLIC signature reaction may serve several compounds)
rxn2comp = {}
for _, r in kid.iterrows():
for x in r['rxn_set']:
if x in cat_rxns:
rxn2comp.setdefault(x, []).append(r['compound_id'])
# carriage restricted to CATABOLIC signature reactions
csig = carr[carr['rxn'].isin(cat_rxns)].copy()
comp_meta = kid.set_index('compound_id')
cat_tot = {cid: sum(1 for x in comp_meta.loc[cid, 'rxn_set'] if x in cat_rxns)
for cid in comp_meta.index}
w = {x: -math.log10(prev_map[x]) for x in cat_rxns}
from collections import defaultdict
acc = defaultdict(lambda: {'rxns': [], 'score': 0.0})
for _, row in csig.iterrows():
rx, gid = row['rxn'], row['genome_id']
for cid in rxn2comp.get(rx, []):
a = acc[(cid, gid)]
a['rxns'].append(rx); a['score'] += w[rx]
pred = []
for (cid, gid), a in acc.items():
m = comp_meta.loc[cid]
carried = sorted(set(a['rxns']))
nsig = len(carried)
# reaction-level tier: degradation-map context beats allowlist-only
tier = 'T2_pathway' if any(x in degmap_rxns for x in carried) else 'T3_signature'
pred.append(dict(compound_id=cid, name=m['name'], npc_pathway=m['npc_pathway'],
kegg_id=m['kegg_id'], genome_id=gid,
tier=tier, n_sig_carried=nsig,
sig_completeness=round(nsig / max(cat_tot[cid], 1), 3),
score=round(a['score'], 3),
sig_rxns_carried=';'.join(carried)))
pred = pd.DataFrame(pred).merge(gmap, on='genome_id', how='left')
pred = pred.sort_values(['compound_id', 'score'], ascending=[True, False]).reset_index(drop=True)
print('prediction rows (compound x genome):', len(pred))
print('distinct compounds with calls :', pred['compound_id'].nunique())
print('distinct genomes called :', pred['genome_id'].nunique())
print(pred['tier'].value_counts().to_string())
prediction rows (compound x genome): 948 distinct compounds with calls : 8 distinct genomes called : 800 tier T2_pathway 846 T3_signature 102
In [8]:
# per-compound summary: how many genomes, tier, top isolate
summ = (pred.groupby(['compound_id', 'name', 'npc_pathway', 'tier'])
.agg(n_genomes=('genome_id', 'nunique'),
max_score=('score', 'max'),
max_completeness=('sig_completeness', 'max'))
.reset_index().sort_values('n_genomes', ascending=False))
print(summ.to_string(index=False))
compound_id name npc_pathway tier n_genomes max_score max_completeness
Cc1_10 3-hydroxybenzoic acid Shikimates and Phenylpropanoids T2_pathway 296 2.669 1.000
Cc1_38 salicylic acid Shikimates and Phenylpropanoids T2_pathway 197 3.035 1.000
Cc1_112 4-hydroxybenzaldehyde Shikimates and Phenylpropanoids T2_pathway 175 3.367 1.000
KKEYFWRCBNTPAC-UHFFFAOYSA-N TEREPHTHALIC ACID Shikimates and Phenylpropanoids T2_pathway 97 1.506 1.000
XNGIFLGASWRNHJ-UHFFFAOYSA-N phthalic acid Shikimates and Phenylpropanoids T2_pathway 81 3.928 0.667
Cc1_208 Phenylethylamine Alkaloids T3_signature 81 1.584 1.000
Cc1_9 xanthine Alkaloids T3_signature 19 2.214 1.000
Cc1_117 Abscisic acid Terpenoids T3_signature 2 3.192 1.000
Figure: candidate-utilizer breadth per compound¶
In [9]:
fig, ax = plt.subplots(figsize=(8, 5))
s = summ.sort_values('n_genomes')
colors = {'T2_pathway': '#2c7fb8', 'T3_signature': '#7fcdbb'}
ax.barh(s['name'], s['n_genomes'], color=[colors[t] for t in s['tier']])
ax.set_xlabel('ENIGMA isolate genomes carrying a signature reaction')
ax.set_title('Reaction-bridge organism calls (n=%d compounds of 83)' % summ['compound_id'].nunique())
from matplotlib.patches import Patch
ax.legend(handles=[Patch(color=colors['T2_pathway'], label='T2 (degradation-context)'),
Patch(color=colors['T3_signature'], label='T3 (signature enzyme)')],
loc='lower right')
fig.tight_layout(); fig.savefig(FIG / '03_organism_calls.png', dpi=150)
print('saved 03_organism_calls.png'); plt.close(fig)
saved 03_organism_calls.png
Save predictions + organism-dark gap list¶
In [10]:
pred_cols = ['compound_id', 'name', 'npc_pathway', 'kegg_id', 'tier',
'genome_id', 'genome_name', 'strain_full', 'ncbi_taxid', 'rank', 'taxon_name',
'n_sig_carried', 'sig_completeness', 'score', 'sig_rxns_carried']
pred[pred_cols].to_csv(DATA / 'compound_organism_predictions.tsv', sep='\t', index=False)
print('wrote data/compound_organism_predictions.tsv', pred.shape)
# compounds with NO enzyme-level organism call — the gap that widens at the organism step
called = set(pred['compound_id'])
dark = link[~link['compound_id'].isin(called)][['compound_id', 'name', 'npc_pathway', 'kegg_id', 'best_tier']].copy()
def why(r):
if pd.isna(r['kegg_id']): return 'no_kegg'
sub = cov[cov.compound_id == r['compound_id']]
if len(sub) == 0: return 'no_kegg'
s = sub.iloc[0]
if s['n_in_gd'] == 0: return 'kegg_no_rxn_in_genomes'
if s['n_sig'] == 0: return 'only_generic_rxns'
return 'only_biosynthetic_signatures' # had signature(s), none catabolic
dark['organism_dark_reason'] = dark.apply(why, axis=1)
dark.to_csv(DATA / 'compound_organism_dark.tsv', sep='\t', index=False)
print('organism-dark compounds:', len(dark), '/', len(link))
print(dark['organism_dark_reason'].value_counts().to_string())
print()
print(dark[['name', 'npc_pathway', 'organism_dark_reason']].to_string(index=False))
wrote data/compound_organism_predictions.tsv (948, 15)
organism-dark compounds: 75 / 83
organism_dark_reason
kegg_no_rxn_in_genomes 33
no_kegg 29
only_biosynthetic_signatures 7
only_generic_rxns 6
name npc_pathway organism_dark_reason
harman Alkaloids kegg_no_rxn_in_genomes
1_prop_2_en_1_yl__1H_indole_3_carboxylic_acid Alkaloids no_kegg
2-hydroxy-4,7,8-trimethylquinoline Alkaloids no_kegg
3,6-dimethylchromone Polyketides no_kegg
7-hydroxy-4,8-dimethylchromen-2-one Shikimates and Phenylpropanoids no_kegg
Fraxetin Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
2',5'-dihydroxychalcone Shikimates and Phenylpropanoids no_kegg
Phthalic acid mono-2-ethylhexyl ester Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
2-Isopropyl-5-methylcyclohexanamine Terpenoids no_kegg
Manool Terpenoids no_kegg
Mellein / Ochracin Shikimates and Phenylpropanoids no_kegg
(-)-Perillyl alcohol Terpenoids kegg_no_rxn_in_genomes
Tyramine Alkaloids only_biosynthetic_signatures
choline sulfate / choline o-sulfuric acid Amino acids and Peptides kegg_no_rxn_in_genomes
5,6-Dimethylbenzimidazole Alkaloids only_generic_rxns
acridone Alkaloids kegg_no_rxn_in_genomes
BIOTIN Alkaloids only_generic_rxns
Triacetonamine Alkaloids no_kegg
Cadaverine Alkaloids only_generic_rxns
4,6,8-trimethylhydroquinolin-2-one Alkaloids no_kegg
Indole-3-butyric acid Alkaloids kegg_no_rxn_in_genomes
indole-3-acetonitrile Alkaloids only_generic_rxns
4-hydroxy-1-methyl-2-quinolone Alkaloids no_kegg
1,4-dimethylquinolin-2(1h)-one Alkaloids no_kegg
2-hydroxyquinoline Alkaloids kegg_no_rxn_in_genomes
3-Indoleacrylic acid Alkaloids, Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
GUANIDINEACETIC ACID Amino acids and Peptides only_biosynthetic_signatures
DODECANEDIOIC ACID Fatty acids kegg_no_rxn_in_genomes
TETRADECANEDIOIC ACID Fatty acids no_kegg
Decanedioic acid/Sebacic acid Fatty acids kegg_no_rxn_in_genomes
Brassylic acid Fatty acids no_kegg
Hexadecanedioic acid Fatty acids kegg_no_rxn_in_genomes
alpha-hydroxyisobutyric acid Fatty acids kegg_no_rxn_in_genomes
12-Hydroxydodecanoic acid Fatty acids kegg_no_rxn_in_genomes
sn-glycero-3-phosphocholine Fatty acids kegg_no_rxn_in_genomes
6-methylchromone Polyketides kegg_no_rxn_in_genomes
3-hydroxybenzyl alcohol Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
4-(4-Hydroxyphenyl)butan-2-one Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
zingerone Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
3-(2-Hydroxyphenyl)propionic acid Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
6-aminochromen-2-one Shikimates and Phenylpropanoids no_kegg
2-Hydroxychalcone Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
7-hydroxyflavanone Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
2-Hydroxyphenylacetic acid Shikimates and Phenylpropanoids only_generic_rxns
guaiacol Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
Cinnamic acid Shikimates and Phenylpropanoids only_biosynthetic_signatures
caffeic acid Shikimates and Phenylpropanoids only_biosynthetic_signatures
ANETHOLE Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
Benzylideneacetone Shikimates and Phenylpropanoids kegg_no_rxn_in_genomes
(±)-Dihydroactinidiolide Terpenoids no_kegg
(S)-(-)-Perillic acid Terpenoids kegg_no_rxn_in_genomes
CARVEOL Terpenoids kegg_no_rxn_in_genomes
(1R)-(-)-Nopol Terpenoids no_kegg
ketopinic acid Terpenoids no_kegg
(+)-Nootkatone Terpenoids kegg_no_rxn_in_genomes
lumichrome Alkaloids no_kegg
2,3-dihydro-1H-indole-7-carboxylic acid Alkaloids no_kegg
INDOLE-3-CARBOXALDEHYDE Alkaloids kegg_no_rxn_in_genomes
6-Hydroxyquinoline Alkaloids no_kegg
(2S)indoline-2-carboxylic acid Alkaloids no_kegg
Palmitic Acid Fatty acids only_biosynthetic_signatures
lauric acid Fatty acids only_biosynthetic_signatures
3-Hydroxyphenylacetic acid Shikimates and Phenylpropanoids only_generic_rxns
Farnesol Terpenoids only_biosynthetic_signatures
D-camphor Terpenoids kegg_no_rxn_in_genomes
NEROLIDOL Terpenoids no_kegg
Bisabolol Terpenoids no_kegg
(-)-Isolongifolol Terpenoids no_kegg
b-Caryophyllene oxide Terpenoids no_kegg
Sclareol Terpenoids kegg_no_rxn_in_genomes
SALSOLINOL Alkaloids kegg_no_rxn_in_genomes
1H-pyrrolo[3,2-b]pyridine-5-carboxylic acid Alkaloids no_kegg
2,6-Pyridinedicarboxylic acid Alkaloids no_kegg
N-methylanthranilic acid Alkaloids kegg_no_rxn_in_genomes
4-Methoxycinnamic acid Shikimates and Phenylpropanoids no_kegg