◆ Microbial Discovery Forge AI Co-Scientist

• Co-Scientist
• Observatory
  • Projects
  • Research Areas
  • Collections
  • Data Explorer
  • Discoveries
  • Community
• Knowledge
  • Skills
  • Memory: Pitfalls
  • Memory: Performance
  • Research Ideas
• About

Log in with ORCiD

Projects / Pan-Bacterial Metal Fitness Atlas / 03_metal_conservation_analysis.ipynb

03 Metal Conservation Analysis

Jupyter notebook from the Pan-Bacterial Metal Fitness Atlas project.

Back to Project

NB 03: Metal Fitness × Pangenome Conservation

Test the two-tier hypothesis: are metal-important genes enriched in the accessory genome (for toxic metals) or the core genome (for essential metals)?

Runs locally — joins NB02 output with existing FB-pangenome link data.

Inputs:

• data/metal_fitness_scores.csv (from NB02)
• conservation_vs_fitness/data/fb_pangenome_link.tsv

Outputs:

• data/metal_conservation_stats.csv
• figures/core_fraction_by_metal.png
• figures/metal_conservation_by_organism.png

import pandas as pd
import numpy as np
from pathlib import Path
from scipy import stats
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
import seaborn as sns

PROJECT_DIR = Path('..').resolve()
DATA_DIR = PROJECT_DIR / 'data'
FIGURES_DIR = PROJECT_DIR / 'figures'

# Load metal fitness scores from NB02
metal_fitness = pd.read_csv(DATA_DIR / 'metal_fitness_scores.csv')
print(f'Metal fitness records: {len(metal_fitness):,}')

# Load FB-pangenome link
fb_link = pd.read_csv(
    PROJECT_DIR.parent / 'conservation_vs_fitness' / 'data' / 'fb_pangenome_link.tsv',
    sep='\t'
)
# Ensure locusId types match
fb_link['locusId'] = fb_link['locusId'].astype(str)
metal_fitness['locusId'] = metal_fitness['locusId'].astype(str)
print(f'FB-pangenome links: {len(fb_link):,}')
print(f'Organisms in link: {fb_link["orgId"].nunique()}')

Metal fitness records: 383,349

FB-pangenome links: 177,863
Organisms in link: 44

1. Join Metal Fitness with Conservation Status

# Join metal fitness with conservation
# Keep only the conservation columns we need
link_cols = ['orgId', 'locusId', 'is_core', 'is_auxiliary', 'is_singleton']
# Some genes map to multiple clades; take the first (most are consistent)
fb_link_dedup = fb_link[link_cols].drop_duplicates(subset=['orgId', 'locusId'])

metal_cons = metal_fitness.merge(
    fb_link_dedup,
    on=['orgId', 'locusId'],
    how='left'
)

n_mapped = metal_cons['is_core'].notna().sum()
print(f'Metal fitness records with conservation data: {n_mapped:,} / {len(metal_cons):,} '
      f'({100*n_mapped/len(metal_cons):.1f}%)')

# Filter to mapped genes only
metal_cons_mapped = metal_cons[metal_cons['is_core'].notna()].copy()
metal_cons_mapped['is_core'] = metal_cons_mapped['is_core'].astype(int)
metal_cons_mapped['is_auxiliary'] = metal_cons_mapped['is_auxiliary'].astype(int)
metal_cons_mapped['is_singleton'] = metal_cons_mapped['is_singleton'].astype(int)

print(f'\nMapped records: {len(metal_cons_mapped):,}')
print(f'Organisms: {metal_cons_mapped["orgId"].nunique()}')
print(f'Metals: {metal_cons_mapped["metal_element"].nunique()}')

Metal fitness records with conservation data: 318,751 / 383,349 (83.1%)

Mapped records: 318,751

Organisms: 22
Metals: 14

2. Core Fraction: Metal-Important vs All Genes

# Overall core fraction
baseline_core = metal_cons_mapped['is_core'].mean()
important_core = metal_cons_mapped.loc[
    metal_cons_mapped['is_metal_important'], 'is_core'
].mean()
not_important_core = metal_cons_mapped.loc[
    ~metal_cons_mapped['is_metal_important'], 'is_core'
].mean()

print(f'Core fraction (all mapped genes):     {baseline_core:.3f}')
print(f'Core fraction (metal-important):      {important_core:.3f}')
print(f'Core fraction (not metal-important):  {not_important_core:.3f}')
print(f'\nDifference: {important_core - not_important_core:+.3f}')

# Fisher exact test: metal-important vs core
imp = metal_cons_mapped[metal_cons_mapped['is_metal_important']]
not_imp = metal_cons_mapped[~metal_cons_mapped['is_metal_important']]
table = [
    [imp['is_core'].sum(), len(imp) - imp['is_core'].sum()],
    [not_imp['is_core'].sum(), len(not_imp) - not_imp['is_core'].sum()]
]
odds_ratio, p_value = stats.fisher_exact(table)
print(f'\nFisher exact test (metal-important vs core):')
print(f'  Odds ratio: {odds_ratio:.3f}')
print(f'  P-value: {p_value:.2e}')
if odds_ratio < 1:
    print(f'  => Metal-important genes are LESS likely to be core (accessory-enriched)')
else:
    print(f'  => Metal-important genes are MORE likely to be core')

Core fraction (all mapped genes):     0.773
Core fraction (metal-important):      0.874
Core fraction (not metal-important):  0.769

Difference: +0.105

Fisher exact test (metal-important vs core):
  Odds ratio: 2.083
  P-value: 4.32e-162
  => Metal-important genes are MORE likely to be core

3. Core Fraction by Metal Type (Toxic vs Essential)

# Add metal category
metal_categories = {
    'Cobalt': 'toxic', 'Nickel': 'toxic', 'Copper': 'toxic',
    'Zinc': 'toxic', 'Aluminum': 'toxic', 'Chromium': 'toxic',
    'Mercury': 'toxic', 'Cadmium': 'toxic', 'Uranium': 'toxic',
    'Iron': 'essential', 'Molybdenum': 'essential', 'Tungsten': 'essential',
    'Selenium': 'essential', 'Manganese': 'essential',
}
metal_cons_mapped['metal_category'] = metal_cons_mapped['metal_element'].map(metal_categories)

# Core fraction per metal, for important genes only
results = []
print('Core fraction of metal-important genes by metal:')
print('=' * 80)

for metal in sorted(metal_cons_mapped['metal_element'].unique()):
    subset = metal_cons_mapped[metal_cons_mapped['metal_element'] == metal]
    imp_subset = subset[subset['is_metal_important']]
    not_imp_subset = subset[~subset['is_metal_important']]
    
    if len(imp_subset) == 0:
        continue
    
    core_imp = imp_subset['is_core'].mean()
    core_baseline = subset['is_core'].mean()
    n_orgs = imp_subset['orgId'].nunique()
    cat = metal_categories.get(metal, '?')
    
    # Fisher test per metal
    if len(not_imp_subset) > 0:
        table = [
            [imp_subset['is_core'].sum(), len(imp_subset) - imp_subset['is_core'].sum()],
            [not_imp_subset['is_core'].sum(), len(not_imp_subset) - not_imp_subset['is_core'].sum()]
        ]
        or_val, p_val = stats.fisher_exact(table)
    else:
        or_val, p_val = np.nan, np.nan
    
    results.append({
        'metal': metal,
        'category': cat,
        'n_important': len(imp_subset),
        'n_organisms': n_orgs,
        'core_frac_important': core_imp,
        'core_frac_baseline': core_baseline,
        'delta': core_imp - core_baseline,
        'odds_ratio': or_val,
        'p_value': p_val,
    })
    
    sig = '*' if p_val < 0.05 else ''
    print(f'  {metal:12s} ({cat:9s})  core={core_imp:.3f}  '
          f'baseline={core_baseline:.3f}  delta={core_imp-core_baseline:+.3f}  '
          f'OR={or_val:.2f}  p={p_val:.2e} {sig}  '
          f'[{len(imp_subset)} genes, {n_orgs} orgs]')

results_df = pd.DataFrame(results)

# Summary by category
print(f'\nSummary by metal category:')
for cat in ['toxic', 'essential']:
    cat_df = results_df[results_df['category'] == cat]
    mean_delta = cat_df['delta'].mean()
    print(f'  {cat:10s}: mean delta = {mean_delta:+.3f} '
          f'({len(cat_df)} metals)')

Core fraction of metal-important genes by metal:
================================================================================
  Aluminum     (toxic    )  core=0.898  baseline=0.791  delta=+0.107  OR=2.37  p=1.19e-26 *  [1381 genes, 14 orgs]
  Cadmium      (toxic    )  core=0.522  baseline=0.531  delta=-0.010  OR=0.96  p=9.16e-01   [92 genes, 1 orgs]
  Chromium     (toxic    )  core=0.710  baseline=0.650  delta=+0.060  OR=1.33  p=3.99e-02 *  [262 genes, 2 orgs]
  Cobalt       (toxic    )  core=0.859  baseline=0.786  delta=+0.072  OR=1.67  p=8.07e-16 *  [1859 genes, 19 orgs]

  Copper       (toxic    )  core=0.867  baseline=0.776  delta=+0.090  OR=1.91  p=3.80e-27 *  [2139 genes, 16 orgs]
  Iron         (essential)  core=0.919  baseline=0.803  delta=+0.116  OR=3.07  p=8.55e-18 *  [651 genes, 3 orgs]
  Manganese    (essential)  core=1.000  baseline=0.802  delta=+0.198  OR=inf  p=2.03e-03 *  [30 genes, 1 orgs]
  Mercury      (toxic    )  core=0.934  baseline=0.802  delta=+0.132  OR=3.62  p=1.61e-04 *  [106 genes, 1 orgs]
  Molybdenum   (essential)  core=0.950  baseline=0.802  delta=+0.148  OR=5.32  p=1.34e-14 *  [302 genes, 1 orgs]
  Nickel       (toxic    )  core=0.877  baseline=0.789  delta=+0.088  OR=1.93  p=3.01e-22 *  [1760 genes, 18 orgs]
  Selenium     (essential)  core=0.933  baseline=0.802  delta=+0.131  OR=3.59  p=2.88e-05 *  [134 genes, 1 orgs]
  Tungsten     (essential)  core=0.947  baseline=0.802  delta=+0.145  OR=4.98  p=5.32e-14 *  [303 genes, 1 orgs]
  Uranium      (toxic    )  core=0.685  baseline=0.650  delta=+0.035  OR=1.18  p=3.39e-01   [178 genes, 2 orgs]
  Zinc         (toxic    )  core=0.890  baseline=0.739  delta=+0.151  OR=2.94  p=2.02e-49 *  [1517 genes, 12 orgs]

Summary by metal category:
  toxic     : mean delta = +0.081 (9 metals)
  essential : mean delta = +0.148 (5 metals)

4. Per-Organism Analysis

# Core fraction of metal-important genes per organism
org_results = []
print('Per-organism: core fraction of metal-important vs all genes')
print('=' * 80)

for org_id in sorted(metal_cons_mapped['orgId'].unique()):
    org_data = metal_cons_mapped[metal_cons_mapped['orgId'] == org_id]
    # Deduplicate: a gene may appear for multiple metals; for organism-level,
    # we want unique genes
    org_genes = org_data.drop_duplicates(subset='locusId')
    org_important = org_data[org_data['is_metal_important']].drop_duplicates(subset='locusId')
    
    core_all = org_genes['is_core'].mean()
    core_imp = org_important['is_core'].mean() if len(org_important) > 0 else np.nan
    
    if len(org_important) > 0 and len(org_genes) > len(org_important):
        org_not_imp = org_genes[~org_genes['locusId'].isin(org_important['locusId'])]
        table = [
            [org_important['is_core'].sum(), len(org_important) - org_important['is_core'].sum()],
            [org_not_imp['is_core'].sum(), len(org_not_imp) - org_not_imp['is_core'].sum()]
        ]
        or_val, p_val = stats.fisher_exact(table)
    else:
        or_val, p_val = np.nan, np.nan
    
    n_metals = org_data['metal_element'].nunique()
    org_results.append({
        'orgId': org_id,
        'n_genes': len(org_genes),
        'n_important': len(org_important),
        'n_metals': n_metals,
        'core_frac_all': core_all,
        'core_frac_important': core_imp,
        'delta': core_imp - core_all if not np.isnan(core_imp) else np.nan,
        'odds_ratio': or_val,
        'p_value': p_val,
    })
    
    sig = '*' if p_val < 0.05 else '' if not np.isnan(p_val) else ''
    print(f'  {org_id:25s}  core_imp={core_imp:.3f}  core_all={core_all:.3f}  '
          f'delta={core_imp-core_all:+.3f}  OR={or_val:.2f}  p={p_val:.2e} {sig}')

org_results_df = pd.DataFrame(org_results)

# How many organisms show metal-important genes as LESS core?
n_less_core = (org_results_df['delta'] < 0).sum()
n_more_core = (org_results_df['delta'] > 0).sum()
n_sig = (org_results_df['p_value'] < 0.05).sum()
print(f'\n{n_less_core}/{len(org_results_df)} organisms: metal-important genes are LESS core')
print(f'{n_more_core}/{len(org_results_df)} organisms: metal-important genes are MORE core')
print(f'{n_sig}/{len(org_results_df)} significant at p<0.05')

Per-organism: core fraction of metal-important vs all genes
================================================================================
  BFirm                      core_imp=0.917  core_all=0.756  delta=+0.161  OR=3.68  p=7.76e-08 *
  Btheta                     core_imp=0.897  core_all=0.633  delta=+0.265  OR=5.51  p=1.93e-24 *
  Caulo                      core_imp=0.918  core_all=0.911  delta=+0.007  OR=1.11  p=6.55e-01 
  Cup4G11                    core_imp=0.761  core_all=0.630  delta=+0.131  OR=1.97  p=2.28e-11 *
  DvH                        core_imp=0.929  core_all=0.802  delta=+0.127  OR=3.82  p=7.61e-17 *

  Korea                      core_imp=0.887  core_all=0.840  delta=+0.047  OR=1.52  p=1.12e-01 
  Koxy                       core_imp=0.885  core_all=0.833  delta=+0.052  OR=1.59  p=9.23e-03 *

  Marino                     core_imp=0.842  core_all=0.730  delta=+0.112  OR=2.07  p=1.91e-06 *
  Methanococcus_JJ           core_imp=0.936  core_all=0.809  delta=+0.127  OR=4.70  p=3.16e-16 *
  Methanococcus_S2           core_imp=0.896  core_all=0.797  delta=+0.099  OR=2.55  p=5.80e-06 *
  Pedo557                    core_imp=0.911  core_all=0.875  delta=+0.035  OR=1.52  p=3.35e-01 
  Phaeo                      core_imp=0.984  core_all=0.851  delta=+0.132  OR=11.13  p=1.05e-09 *
  Ponti                      core_imp=1.000  core_all=0.997  delta=+0.003  OR=inf  p=6.20e-01 
  SynE                       core_imp=0.993  core_all=0.980  delta=+0.013  OR=3.37  p=2.95e-02 *
  WCS417                     core_imp=0.978  core_all=0.921  delta=+0.057  OR=3.87  p=4.59e-02 *

  acidovorax_3H11            core_imp=0.789  core_all=0.748  delta=+0.041  OR=1.29  p=9.56e-02 
  psRCH2                     core_imp=0.571  core_all=0.531  delta=+0.040  OR=1.20  p=6.93e-02 
  pseudo13_GW456_L13         core_imp=0.972  core_all=0.832  delta=+0.140  OR=7.35  p=1.32e-07 *
  pseudo1_N1B4               core_imp=0.755  core_all=0.741  delta=+0.015  OR=1.08  p=7.30e-01 
  pseudo3_N2E3               core_imp=0.972  core_all=0.975  delta=-0.003  OR=0.89  p=6.96e-01 
  pseudo5_N2C3_1             core_imp=0.798  core_all=0.684  delta=+0.114  OR=1.90  p=7.24e-06 *
  pseudo6_N2E2               core_imp=0.944  core_all=0.798  delta=+0.146  OR=4.49  p=8.69e-10 *

1/22 organisms: metal-important genes are LESS core
21/22 organisms: metal-important genes are MORE core
14/22 significant at p<0.05

5. Figures

# Bar chart: core fraction by metal (important vs baseline)
fig, ax = plt.subplots(figsize=(14, 6))

# Sort by delta
plot_df = results_df.sort_values('delta')
x = range(len(plot_df))
colors = ['#e74c3c' if c == 'toxic' else '#3498db' for c in plot_df['category']]

bars = ax.bar(x, plot_df['core_frac_important'], color=colors, alpha=0.8, edgecolor='black', linewidth=0.5)
ax.scatter(x, plot_df['core_frac_baseline'], color='black', marker='_', s=200, linewidths=2, zorder=5, label='Baseline (all genes)')

ax.set_xticks(x)
ax.set_xticklabels(plot_df['metal'], rotation=45, ha='right')
ax.set_ylabel('Fraction Core Genome')
ax.set_title('Core Genome Fraction: Metal-Important Genes vs Baseline', fontsize=14)
ax.legend()

# Add significance stars
for i, (_, row) in enumerate(plot_df.iterrows()):
    if row['p_value'] < 0.001:
        ax.text(i, row['core_frac_important'] + 0.01, '***', ha='center', fontsize=10)
    elif row['p_value'] < 0.01:
        ax.text(i, row['core_frac_important'] + 0.01, '**', ha='center', fontsize=10)
    elif row['p_value'] < 0.05:
        ax.text(i, row['core_frac_important'] + 0.01, '*', ha='center', fontsize=10)

# Legend for colors
from matplotlib.patches import Patch
legend_elements = [
    Patch(facecolor='#e74c3c', alpha=0.8, label='Toxic metal'),
    Patch(facecolor='#3498db', alpha=0.8, label='Essential metal'),
    plt.Line2D([0], [0], marker='_', color='black', markersize=10, linestyle='None', label='Baseline'),
]
ax.legend(handles=legend_elements, loc='lower right')

plt.tight_layout()
fig.savefig(FIGURES_DIR / 'core_fraction_by_metal.png', dpi=150, bbox_inches='tight')
plt.show()
print(f'Saved: figures/core_fraction_by_metal.png')

Saved: figures/core_fraction_by_metal.png

# Per-organism delta plot
fig, ax = plt.subplots(figsize=(12, 6))
plot_org = org_results_df.sort_values('delta')
colors = ['#e74c3c' if d < 0 else '#2ecc71' for d in plot_org['delta']]
sig_markers = ['*' if p < 0.05 else '' for p in plot_org['p_value']]

ax.barh(range(len(plot_org)), plot_org['delta'], color=colors, alpha=0.8, edgecolor='black', linewidth=0.5)
ax.set_yticks(range(len(plot_org)))
ax.set_yticklabels(plot_org['orgId'], fontsize=9)
ax.axvline(0, color='black', linestyle='-', linewidth=0.5)
ax.set_xlabel('Delta (core fraction: metal-important - all genes)')
ax.set_title('Metal-Important Genes: Core Enrichment by Organism', fontsize=14)

for i, (_, row) in enumerate(plot_org.iterrows()):
    if row['p_value'] < 0.05:
        ax.text(row['delta'] + 0.002 * np.sign(row['delta']), i, '*', 
                ha='center', va='center', fontsize=12, fontweight='bold')

plt.tight_layout()
fig.savefig(FIGURES_DIR / 'metal_conservation_by_organism.png', dpi=150, bbox_inches='tight')
plt.show()
print(f'Saved: figures/metal_conservation_by_organism.png')

Saved: figures/metal_conservation_by_organism.png

6. Save Results

# --- Issue #3: Formal toxic vs essential comparison ---
from scipy.stats import mannwhitneyu

toxic_deltas = results_df[results_df['category'] == 'toxic']['delta'].values
essential_deltas = results_df[results_df['category'] == 'essential']['delta'].values

u_stat, mw_p = mannwhitneyu(essential_deltas, toxic_deltas, alternative='greater')
print('Formal comparison: Essential vs Toxic metal core enrichment')
print(f'  Essential mean delta: {essential_deltas.mean():+.3f} (n={len(essential_deltas)})')
print(f'  Toxic mean delta:    {toxic_deltas.mean():+.3f} (n={len(toxic_deltas)})')
print(f'  Mann-Whitney U: {u_stat:.1f}, p={mw_p:.4f} (one-sided: essential > toxic)')
print()

# --- Issue #7: Phylogenetic sensitivity analysis ---
# Exclude duplicate Pseudomonas fluorescens strains (keep only pseudo3_N2E3 as representative)
pseudo_strains = ['pseudo1_N1B4', 'pseudo5_N2C3_1', 'pseudo6_N2E2', 'pseudo13_GW456_L13']
metal_cons_no_pseudo = metal_cons_mapped[~metal_cons_mapped['orgId'].isin(pseudo_strains)]

imp_np = metal_cons_no_pseudo[metal_cons_no_pseudo['is_metal_important']]
not_imp_np = metal_cons_no_pseudo[~metal_cons_no_pseudo['is_metal_important']]

core_imp_np = imp_np['is_core'].mean()
core_not_np = not_imp_np['is_core'].mean()
table_np = [
    [imp_np['is_core'].sum(), len(imp_np) - imp_np['is_core'].sum()],
    [not_imp_np['is_core'].sum(), len(not_imp_np) - not_imp_np['is_core'].sum()]
]
or_np, p_np = stats.fisher_exact(table_np)

print('Phylogenetic sensitivity: excluding 4 duplicate P. fluorescens strains')
print(f'  Organisms: {metal_cons_no_pseudo["orgId"].nunique()} (was {metal_cons_mapped["orgId"].nunique()})')
print(f'  Core fraction (metal-important): {core_imp_np:.3f} (was {important_core:.3f})')
print(f'  Core fraction (not important):   {core_not_np:.3f} (was {not_important_core:.3f})')
print(f'  OR={or_np:.3f}, p={p_np:.2e}')
print(f'  Result: {"ROBUST" if or_np > 1 and p_np < 0.05 else "NOT ROBUST"} — '
      f'core enrichment {"persists" if or_np > 1 else "lost"} after removing Pseudomonas duplicates')
print()

# Save results
results_df.to_csv(DATA_DIR / 'metal_conservation_stats.csv', index=False)
org_results_df.to_csv(DATA_DIR / 'organism_conservation_stats.csv', index=False)

# Save sensitivity analysis results
sensitivity = pd.DataFrame([{
    'analysis': 'full', 'n_organisms': metal_cons_mapped['orgId'].nunique(),
    'core_imp': important_core, 'core_not': not_important_core,
    'OR': odds_ratio, 'p_value': p_value,
}, {
    'analysis': 'no_pseudo_duplicates', 'n_organisms': metal_cons_no_pseudo['orgId'].nunique(),
    'core_imp': core_imp_np, 'core_not': core_not_np,
    'OR': or_np, 'p_value': p_np,
}])
sensitivity.to_csv(DATA_DIR / 'sensitivity_analysis.csv', index=False)

print(f'Saved: data/metal_conservation_stats.csv ({len(results_df)} metals)')
print(f'Saved: data/organism_conservation_stats.csv ({len(org_results_df)} organisms)')
print(f'Saved: data/sensitivity_analysis.csv')

print('\n' + '=' * 80)
print('NB03 SUMMARY: Metal Fitness x Pangenome Conservation')
print('=' * 80)
print(f'Mapped gene-metal records: {len(metal_cons_mapped):,}')
print(f'Overall: metal-important genes are {"LESS" if important_core < not_important_core else "MORE"} '
      f'likely core (OR={odds_ratio:.3f}, p={p_value:.2e})')
print(f'Toxic metals mean delta: {results_df[results_df["category"]=="toxic"]["delta"].mean():+.3f}')
print(f'Essential metals mean delta: {results_df[results_df["category"]=="essential"]["delta"].mean():+.3f}')
print(f'Essential > Toxic: Mann-Whitney p={mw_p:.4f}')
print(f'{n_less_core}/{len(org_results_df)} organisms show metal genes as less core')
print(f'{n_sig}/{len(org_results_df)} significant at p<0.05')
print(f'Sensitivity (excl. Pseudomonas duplicates): OR={or_np:.3f}, p={p_np:.2e} — ROBUST')
print('=' * 80)

Formal comparison: Essential vs Toxic metal core enrichment
  Essential mean delta: +0.148 (n=5)
  Toxic mean delta:    +0.081 (n=9)
  Mann-Whitney U: 39.0, p=0.0145 (one-sided: essential > toxic)

Phylogenetic sensitivity: excluding 4 duplicate P. fluorescens strains
  Organisms: 18 (was 22)
  Core fraction (metal-important): 0.875 (was 0.874)
  Core fraction (not important):   0.772 (was 0.769)
  OR=2.065, p=5.86e-141
  Result: ROBUST — core enrichment persists after removing Pseudomonas duplicates

Saved: data/metal_conservation_stats.csv (14 metals)
Saved: data/organism_conservation_stats.csv (22 organisms)
Saved: data/sensitivity_analysis.csv

================================================================================
NB03 SUMMARY: Metal Fitness x Pangenome Conservation
================================================================================
Mapped gene-metal records: 318,751
Overall: metal-important genes are MORE likely core (OR=2.083, p=4.32e-162)
Toxic metals mean delta: +0.081
Essential metals mean delta: +0.148
Essential > Toxic: Mann-Whitney p=0.0145
1/22 organisms show metal genes as less core
14/22 significant at p<0.05
Sensitivity (excl. Pseudomonas duplicates): OR=2.065, p=5.86e-141 — ROBUST
================================================================================

Microbial Discovery Forge

AI co-scientist and research observatory for BERDL-scale microbial discovery.

Maintained by Paramvir S. Dehal

Observatory

• Projects
• Collections
• Discoveries

Knowledge

• Skills
• Pitfalls
• Performance
• Research Ideas

Resources

• BERDL JupyterHub
• Co-Scientist
• How to Cite

Built on BERDL (KBase). Last updated: 2026-04-03 21:43 UTC

×