01 Leaden2018 Fur Signature
Jupyter notebook from the Caulobacter Fur–Lipid A Loss project.
01 — Leaden 2018 Δfur signature: clean Fur-only DEG reference¶
Project: caulobacter_fur_lipida_loss — Phase C, NB01.
Purpose¶
Disambiguate Fur-specific vs SspB-specific effects in the colleague's 4584-vs-4580 contrast (which is Δfur ΔsspB ΔrsaA on a ΔrsaA baseline) by referencing it against Leaden et al. 2018 (PMID 30210482), which compared WT vs Δfur under iron-replete conditions plus WT under iron limitation (2h DP).
Preflight succeeded — the published supplementary Table 2.XLSX is at ~/data/kr-caulobacter-envelope/raw/Table 2.XLSX; CTS re-analysis from raw SRA is not required. This saves ~3-6 h cluster compute.
Stop condition (per RESEARCH_PLAN v2)¶
Spearman concordance between Leaden Δfur log2-FC and 4584-vs-4580 log2-FC, over the genes that are differentially expressed in Leaden (≈99 genes with logFC reported):
- ≥ 0.30: pass — Δfur is a major driver of 4584-vs-4580; H2's "critical Fur regulon subset" framing is intact.
- < 0.30: fail — ΔsspB dominates the 4584-vs-4580 signal; trigger plan reframe at the checkpoint.
A negative or near-zero concordance would suggest the two perturbations push different directions on the same genes — an anti-correspondence, also a plan-reframe trigger.
Outputs¶
data/NB01_leaden_fur_de.csv— Leaden Δfur DEGs in tidy formdata/NB01_leaden_iron_de.csv— Leaden iron-limitation (2h DP) DEGsdata/NB01_fur_only_signature.csv— H2 candidate Fur-released genes intersected with 4584-vs-4580data/NB01_source_provenance.md— record that preflight succeededfigures/NB01_fur_signature_scatter.png— Leaden Δfur logFC vs 4584-vs-4580 logFC scatterfigures/NB01_leaden_iron_vs_fur.png— Leaden's own iron-limitation vs Δfur logFC (internal QC)
In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from pathlib import Path
from scipy.stats import spearmanr, pearsonr
import openpyxl
pd.set_option('display.max_colwidth', 80)
pd.set_option('display.width', 180)
sns.set_context('notebook')
sns.set_style('whitegrid')
PROJ = Path('/home/aparkin/BERIL-research-observatory/projects/caulobacter_fur_lipida_loss')
USERDATA = Path('/home/aparkin/data/kr-caulobacter-envelope')
DATA_OUT = PROJ / 'data'
FIG = PROJ / 'figures'
LEADEN_PATH = USERDATA / 'raw' / 'Table 2.XLSX'
assert LEADEN_PATH.exists(), f'Missing Leaden Table 2 at {LEADEN_PATH}'
print(f'Leaden source: {LEADEN_PATH}')
# colleague's clean tables
diff = pd.read_csv(USERDATA / 'clean' / 'fact_differential.csv')
feat = pd.read_csv(USERDATA / 'clean' / 'dim_feature.csv')
print(f'diff: {diff.shape}, feat: {feat.shape}')
Leaden source: /home/aparkin/data/kr-caulobacter-envelope/raw/Table 2.XLSX diff: (11871, 8), feat: (4020, 4)
1. Parse Leaden Table 2 — multi-sheet → tidy DataFrames¶
Sheet layout (verified in preflight):
| Sheet | Direction | Source |
|---|---|---|
up wt-DP and fur |
up in BOTH 2h iron-deprivation AND Δfur | both |
down wt-DP and fur |
down in both | both |
up fur |
up in Δfur only | Δfur only |
down fur |
down in Δfur only | Δfur only |
up wt-DP |
up in 2h DP only | iron-limitation only |
down wt-DP |
down in 2h DP only | iron-limitation only |
Each sheet has columns: #, Gene (CCNA locus), Fold change (log2) 2h DP, Fold change (log2) fur mutant (some sheets only have one), Product, Category.
Result: build a long-format DataFrame keyed on (CCNA locus, condition ∈ {fur, iron}) with logFC.
In [2]:
# Load workbook
wb = openpyxl.load_workbook(LEADEN_PATH, data_only=True)
print(f'Sheets: {wb.sheetnames}')
def parse_sheet(ws, sheet_name):
"""Parse one DE sheet to a tidy DataFrame. Returns (df_fur, df_iron) where either may be empty."""
rows = list(ws.iter_rows(values_only=True))
if not rows:
return pd.DataFrame(), pd.DataFrame()
# Find header row (first row containing 'Gene' in any column)
header_idx = None
for i, r in enumerate(rows[:5]):
if r and any((c or '').strip() == 'Gene' for c in r if isinstance(c, str)):
header_idx = i
break
if header_idx is None:
return pd.DataFrame(), pd.DataFrame()
header = [str(c).strip().replace('\n', ' ') if c else '' for c in rows[header_idx]]
data = [r for r in rows[header_idx+1:] if r and r[1] is not None and isinstance(r[1], str) and r[1].startswith('CCNA_')]
df = pd.DataFrame(data, columns=header)
# find logFC columns
fur_col = next((c for c in df.columns if 'fur' in c.lower() and 'fold' in c.lower()), None)
iron_col = next((c for c in df.columns if 'dp' in c.lower() and 'fold' in c.lower()), None)
product_col = next((c for c in df.columns if c.strip() == 'Product'), None)
cat_col = next((c for c in df.columns if c.strip() == 'Category'), None)
df_fur = pd.DataFrame()
df_iron = pd.DataFrame()
if fur_col:
df_fur = pd.DataFrame({
'locustag': df['Gene'].astype(str).str.strip(),
'log2fc_fur': pd.to_numeric(df[fur_col], errors='coerce'),
'product': df[product_col].astype(str).str.strip() if product_col else '',
'category': df[cat_col].astype(str).str.strip() if cat_col else '',
'leaden_sheet': sheet_name,
}).dropna(subset=['log2fc_fur'])
if iron_col:
df_iron = pd.DataFrame({
'locustag': df['Gene'].astype(str).str.strip(),
'log2fc_iron_2hDP': pd.to_numeric(df[iron_col], errors='coerce'),
'product': df[product_col].astype(str).str.strip() if product_col else '',
'category': df[cat_col].astype(str).str.strip() if cat_col else '',
'leaden_sheet': sheet_name,
}).dropna(subset=['log2fc_iron_2hDP'])
return df_fur, df_iron
all_fur = []
all_iron = []
for s in wb.sheetnames:
if s.lower().startswith('summary'):
continue
fur, iron = parse_sheet(wb[s], s)
if not fur.empty: all_fur.append(fur)
if not iron.empty: all_iron.append(iron)
leaden_fur = pd.concat(all_fur, ignore_index=True).drop_duplicates('locustag')
leaden_iron = pd.concat(all_iron, ignore_index=True).drop_duplicates('locustag')
print(f'Leaden Δfur DEGs (deduplicated): {len(leaden_fur)}')
print(f'Leaden iron-limitation DEGs (deduplicated): {len(leaden_iron)}')
# Direction labels (for clarity)
leaden_fur['direction_fur'] = np.sign(leaden_fur['log2fc_fur']).map({1:'up', -1:'down', 0:'none'})
leaden_iron['direction_iron'] = np.sign(leaden_iron['log2fc_iron_2hDP']).map({1:'up', -1:'down', 0:'none'})
print('\nΔfur direction distribution:')
print(leaden_fur['direction_fur'].value_counts())
print('\nLeaden Δfur sheet contributions:')
print(leaden_fur['leaden_sheet'].value_counts())
display(leaden_fur.head())
Sheets: ['Summary', 'up wt-DP and fur ', 'down wt-DP and fur', 'up fur ', 'down fur', 'up wt-DP ', 'down wt-DP'] Leaden Δfur DEGs (deduplicated): 93 Leaden iron-limitation DEGs (deduplicated): 491 Δfur direction distribution: direction_fur down 53 up 40 Name: count, dtype: int64 Leaden Δfur sheet contributions: leaden_sheet down wt-DP and fur 38 up wt-DP and fur 23 up fur 17 down fur 15 Name: count, dtype: int64
| locustag | log2fc_fur | product | category | leaden_sheet | direction_fur | |
|---|---|---|---|---|---|---|
| 0 | CCNA_00039 | 1.738843 | hypothetical protein | hypotetical protein | up wt-DP and fur | up |
| 1 | CCNA_00263 | 1.909522 | hypothetical protein | hypothetical protein | up wt-DP and fur | up |
| 2 | CCNA_03932 | 2.331872 | hypothetical protein | hypothetical protein | up wt-DP and fur | up |
| 3 | CCNA_03933 | 2.474392 | hypothetical protein | hypothetical protein | up wt-DP and fur | up |
| 4 | CCNA_02278 | 1.575633 | hypothetical protein | hypothetical protein | up wt-DP and fur | up |
In [3]:
# Save tidy tables
leaden_fur.to_csv(DATA_OUT / 'NB01_leaden_fur_de.csv', index=False)
leaden_iron.to_csv(DATA_OUT / 'NB01_leaden_iron_de.csv', index=False)
print(f'Saved NB01_leaden_fur_de.csv ({len(leaden_fur)} rows)')
print(f'Saved NB01_leaden_iron_de.csv ({len(leaden_iron)} rows)')
Saved NB01_leaden_fur_de.csv (93 rows) Saved NB01_leaden_iron_de.csv (491 rows)
2. Internal QC — Leaden Δfur vs iron-limitation (2h DP)¶
Leaden's own data give an upper-bound expectation for what "Fur-like" perturbations look like. Δfur and iron-limitation should be correlated but not identical. Genes that are co-regulated (both up, both down) are the core Fur-iron axis. Genes that respond only to one are condition-specific.
In [4]:
# Merge Leaden's own two signatures on locustag
leaden_merged = leaden_fur[['locustag','log2fc_fur','product','category']].merge(
leaden_iron[['locustag','log2fc_iron_2hDP']],
on='locustag', how='outer'
)
both = leaden_merged.dropna(subset=['log2fc_fur','log2fc_iron_2hDP'])
print(f'Genes in BOTH Δfur and iron-limitation DEG lists: {len(both)}')
# Spearman + Pearson over the shared subset
if len(both) >= 5:
rho_s, p_s = spearmanr(both['log2fc_fur'], both['log2fc_iron_2hDP'])
rho_p, p_p = pearsonr(both['log2fc_fur'], both['log2fc_iron_2hDP'])
print(f'\nLeaden internal: Δfur vs iron-limitation logFC')
print(f' Spearman ρ = {rho_s:.3f}, p = {p_s:.2e}')
print(f' Pearson r = {rho_p:.3f}, p = {p_p:.2e}')
# Plot
fig, ax = plt.subplots(figsize=(7, 6))
ax.scatter(both['log2fc_iron_2hDP'], both['log2fc_fur'], alpha=0.7, s=30, edgecolor='k', linewidth=0.3)
ax.axhline(0, color='grey', lw=0.5); ax.axvline(0, color='grey', lw=0.5)
ax.plot([-6,6],[-6,6], color='red', lw=0.5, alpha=0.5, label='y=x')
ax.set_xlabel('Leaden 2018 — log2FC iron-limited 2h DP / WT')
ax.set_ylabel('Leaden 2018 — log2FC Δfur / WT')
ax.set_title(f'Leaden internal: Δfur ⟷ iron-limitation (n={len(both)} shared DEGs)')
ax.legend()
plt.tight_layout()
plt.savefig(FIG / 'NB01_leaden_iron_vs_fur.png', dpi=140, bbox_inches='tight')
plt.show()
Genes in BOTH Δfur and iron-limitation DEG lists: 62 Leaden internal: Δfur vs iron-limitation logFC Spearman ρ = 0.928, p = 2.45e-27 Pearson r = 0.938, p = 2.35e-29
3. Build the join with the colleague's 4584-vs-4580 contrast¶
The colleague's fact_differential.csv has all loci (significant + non-significant) for the 4584-vs-4580 (ΔrsaA Δfur ΔsspB vs ΔrsaA) contrast. Leaden's Table 2 only reports significant DEGs (no logFC for non-DEGs). So the concordance check is over the genes Leaden flagged as significant, looked up in our data.
In [5]:
# 4584-vs-4580 logFCs by locus
ours = diff[diff['contrast']=='4584_vs_4580'][['locustag','logFC','fdr','pvalue']].rename(
columns={'logFC':'log2fc_ours','fdr':'fdr_ours','pvalue':'pvalue_ours'})
# Inner join: only genes in Leaden Δfur list with available data in our 4584-vs-4580
joined = leaden_fur[['locustag','log2fc_fur','product','category','leaden_sheet']].merge(
ours, on='locustag', how='left')
n_total = len(joined)
n_with_ours = joined['log2fc_ours'].notna().sum()
print(f'Leaden Δfur DEGs: {n_total}')
print(f'Of these, found in our DE: {n_with_ours} ({n_with_ours/n_total*100:.0f}%)')
# Concordance over genes present in both
both = joined.dropna(subset=['log2fc_ours','log2fc_fur'])
rho_s, p_s = spearmanr(both['log2fc_fur'], both['log2fc_ours'])
rho_p, p_p = pearsonr(both['log2fc_fur'], both['log2fc_ours'])
# Also check sign concordance (binary): same direction?
sign_concord = (np.sign(both['log2fc_fur']) == np.sign(both['log2fc_ours'])).mean()
print(f'\n=== STOP CONDITION CHECK ===')
print(f'Joined pairs: n = {len(both)}')
print(f'Spearman ρ (Leaden Δfur ↔ ours 4584-vs-4580): {rho_s:.3f}, p = {p_s:.2e}')
print(f'Pearson r: {rho_p:.3f}, p = {p_p:.2e}')
print(f'Sign concordance (same direction): {sign_concord:.1%}')
VERDICT = 'PASS — Δfur is a major driver' if rho_s >= 0.30 else 'FAIL — trigger plan reframe'
print(f'\nVerdict (threshold ρ ≥ 0.30): {VERDICT}')
joined.to_csv(DATA_OUT / 'NB01_fur_only_signature.csv', index=False)
Leaden Δfur DEGs: 93 Of these, found in our DE: 93 (100%) === STOP CONDITION CHECK === Joined pairs: n = 93 Spearman ρ (Leaden Δfur ↔ ours 4584-vs-4580): 0.315, p = 2.08e-03 Pearson r: 0.453, p = 5.27e-06 Sign concordance (same direction): 71.0% Verdict (threshold ρ ≥ 0.30): PASS — Δfur is a major driver
4. Scatter — Leaden Δfur logFC vs 4584-vs-4580 logFC¶
Colored by Leaden category (functional class) where annotated.
In [6]:
fig, ax = plt.subplots(figsize=(9, 7))
plot_df = both.copy()
# Tidy category — coarse-grain
plot_df['cat_short'] = plot_df['category'].fillna('Other').str.strip().str[:30]
top_cats = plot_df['cat_short'].value_counts().head(6).index.tolist()
plot_df['cat_plot'] = plot_df['cat_short'].where(plot_df['cat_short'].isin(top_cats), 'Other')
palette = sns.color_palette('tab10', n_colors=len(plot_df['cat_plot'].unique()))
for (cat, sub), c in zip(plot_df.groupby('cat_plot'), palette):
ax.scatter(sub['log2fc_fur'], sub['log2fc_ours'], label=f'{cat} (n={len(sub)})',
alpha=0.75, s=45, edgecolor='k', linewidth=0.3, color=c)
# y=x line
lim = max(abs(plot_df['log2fc_fur']).max(), abs(plot_df['log2fc_ours']).max()) * 1.1
ax.plot([-lim, lim], [-lim, lim], '--', color='red', alpha=0.4, label='y=x')
ax.axhline(0, color='grey', lw=0.5); ax.axvline(0, color='grey', lw=0.5)
ax.set_xlim(-lim, lim); ax.set_ylim(-lim, lim)
ax.set_xlabel('Leaden 2018 — log2FC Δfur / WT')
ax.set_ylabel('Our data — log2FC 4584 / 4580 (Δfur ΔsspB ΔrsaA vs ΔrsaA)')
ax.set_title(f'Fur-only signature concordance (Spearman ρ={rho_s:.3f}, n={len(both)})')
ax.legend(fontsize=8, loc='lower right')
plt.tight_layout()
plt.savefig(FIG / 'NB01_fur_signature_scatter.png', dpi=140, bbox_inches='tight')
plt.show()
5. Genes diagnosed as Fur-only, SspB-influenced, or divergent¶
Categorize Leaden Δfur DEGs by behavior in our 4584-vs-4580 contrast:
- Concordant strong: same direction, |logFC ours| > 1
- Concordant weak: same direction, |logFC ours| ≤ 1 (likely Fur effect partially masked by SspB)
- Discordant: opposite direction (likely SspB-dominant or context-dependent)
- Not differential in ours: |logFC ours| < 0.5 (Fur effect compensated/buffered in our background)
In [7]:
def categorize(row):
if pd.isna(row['log2fc_ours']):
return 'missing'
same_dir = np.sign(row['log2fc_fur']) == np.sign(row['log2fc_ours'])
abs_ours = abs(row['log2fc_ours'])
if not same_dir and abs_ours > 0.5:
return 'discordant'
if same_dir and abs_ours > 1:
return 'concordant_strong'
if same_dir and abs_ours > 0.5:
return 'concordant_weak'
return 'buffered'
joined['category_call'] = joined.apply(categorize, axis=1)
cat_counts = joined['category_call'].value_counts()
print('Category breakdown for Leaden Δfur DEGs:')
print(cat_counts)
print()
# Show concordant-strong (the core Fur signature)
print('=== Concordant-strong: clean Fur signature in our data ===')
disp_cols = ['locustag','log2fc_fur','log2fc_ours','fdr_ours','product','category']
strong = joined[joined['category_call']=='concordant_strong'].sort_values('log2fc_fur', ascending=False)
display(strong[disp_cols])
print('\n=== Discordant: opposing Δfur and 4584-vs-4580 (SspB-dominant?) ===')
disc = joined[joined['category_call']=='discordant'].sort_values('log2fc_fur', ascending=False)
display(disc[disp_cols])
print('\n=== Buffered: significant in Leaden Δfur, not in our data ===')
buff = joined[joined['category_call']=='buffered'].sort_values('log2fc_fur', ascending=False)
display(buff[disp_cols].head(20))
Category breakdown for Leaden Δfur DEGs: category_call buffered 53 concordant_strong 32 concordant_weak 5 discordant 3 Name: count, dtype: int64 === Concordant-strong: clean Fur signature in our data ===
| locustag | log2fc_fur | log2fc_ours | fdr_ours | product | category | |
|---|---|---|---|---|---|---|
| 6 | CCNA_03904 | 3.292590 | 7.424133 | 9.818341e-09 | hypothetical protein | hypothetical protein |
| 77 | CCNA_R0088 | 2.961138 | 1.221972 | 2.028937e-02 | minimal medium expressed sRNA | non-coding RNA |
| 5 | CCNA_02452 | 2.797598 | 4.263908 | 4.008802e-08 | hypothetical protein | hypothetical protein |
| 8 | CCNA_03372 | 2.481713 | 3.089329 | 7.714849e-04 | bacterioferritin-associated ferredoxin | iron metabolism |
| 12 | CCNA_03155 | 2.450797 | 3.935256 | 3.445656e-05 | PepSY-associated transmembrane protein | Miscellaneous |
| 75 | CCNA_02275 | 2.335534 | 9.847587 | 9.778952e-09 | ABC transporter, periplasmic component | transport |
| 13 | CCNA_03156 | 2.286528 | 3.798423 | 9.233978e-05 | putative periplasmic protein | Miscellaneous |
| 20 | CCNA_02277 | 2.173145 | 10.510821 | 6.676913e-10 | TonB-dependent outer membrane channel | Transport |
| 70 | CCNA_02274 | 2.148901 | 8.265150 | 1.339354e-09 | EF-Hand domain protein | miscellaneous |
| 71 | CCNA_00733 | 2.109850 | 1.838164 | 3.136734e-05 | GumN superfamily protein | miscellaneous |
| 14 | CCNA_03157 | 2.088265 | 3.984614 | 6.816612e-06 | hypothetical protein | Miscellaneous |
| 9 | CCNA_03158 | 1.965915 | 3.806852 | 7.891681e-05 | iron-sulfur cluster assembly/repair protein ApbE | Iron-sulfur cluster assembly/repair |
| 69 | CCNA_02273 | 1.861318 | 1.937235 | 3.328940e-06 | glutathione peroxidase | miscellaneous |
| 74 | CCNA_03159 | 1.832140 | 3.596584 | 1.959441e-04 | sulfite reductase (NADPH) flavoprotein alpha-component | sulfur metabolism |
| 15 | CCNA_03997 | 1.795868 | 1.822529 | 1.521732e-05 | amelogenin/CpxP-related protein | Miscellaneous |
| 68 | CCNA_01443 | 1.770546 | 1.001482 | 7.811350e-03 | hypothetical protein | hypothetical protein |
| 76 | CCNA_03023 | 1.721918 | 2.964392 | 6.436961e-06 | TonB-dependent receptor | transport |
| 21 | CCNA_03022 | 1.708127 | 3.025971 | 1.696068e-05 | PiuB-family transmembrane transporter | Transport |
| 17 | CCNA_00028 | 1.657326 | 4.346785 | 2.078017e-04 | TonB-dependent receptor | Transport |
| 19 | CCNA_00748 | 1.638706 | 1.941247 | 1.735812e-05 | ferrous iron transport protein A | Transport |
| 4 | CCNA_02278 | 1.575633 | 1.403990 | 3.717389e-05 | hypothetical protein | hypothetical protein |
| 67 | CCNA_01090 | 1.555252 | 1.102670 | 4.300214e-05 | hypothetical protein | hypothetical protein |
| 66 | CCNA_01089 | 1.364063 | 1.293789 | 5.513603e-05 | hypothetical protein | hypothetical protein |
| 35 | CCNA_03641 | -1.458736 | -1.507411 | 2.694731e-01 | succinate dehydrogenase iron-sulfur protein | Energy metabolism |
| 33 | CCNA_02031 | -1.768990 | -1.424882 | 4.719260e-05 | NADH-quinone oxidoreductase chain C | Energy metabolism |
| 36 | CCNA_03642 | -1.934064 | -1.514384 | 1.236055e-06 | succinate dehydrogenase flavoprotein subunit | Energy metabolism |
| 90 | CCNA_02910 | -1.992922 | -1.146710 | 1.241171e-02 | TonB-dependent receptor | Transport |
| 37 | CCNA_03644 | -2.509638 | -1.734002 | 3.761314e-05 | succinate dehydrogenase cytochrome B-556 subunit | Energy metabolism |
| 84 | CCNA_00210 | -2.900768 | -1.030624 | 8.567700e-03 | TonB-dependent receptor | Transport |
| 89 | CCNA_02048 | -3.200863 | -1.371913 | 2.190433e-03 | TonB-dependent receptor | Transport |
| 91 | CCNA_03108 | -3.226714 | -1.730037 | 3.143158e-03 | TonB-dependent outer membrane receptor | Transport |
| 83 | CCNA_00055 | -7.121931 | -4.085965 | 4.008802e-08 | ferric uptake regulation protein | Regulator |
=== Discordant: opposing Δfur and 4584-vs-4580 (SspB-dominant?) ===
| locustag | log2fc_fur | log2fc_ours | fdr_ours | product | category | |
|---|---|---|---|---|---|---|
| 7 | CCNA_02999 | 2.539398 | -0.566960 | 0.101232 | hypothetical protein | hypothetical protein |
| 47 | CCNA_01519 | -3.679348 | 2.185417 | 0.000003 | universal stress protein family | Miscellaneous |
| 43 | CCNA_00754 | -6.604926 | 0.602951 | 0.256430 | Nodulin-21 / CCC1-related protein | Miscellaneous |
=== Buffered: significant in Leaden Δfur, not in our data ===
| locustag | log2fc_fur | log2fc_ours | fdr_ours | product | category | |
|---|---|---|---|---|---|---|
| 65 | CCNA_01851 | 3.288361 | -0.411724 | 0.827810 | quinol cytochrome oxidase polypeptide II | energy metabolism |
| 62 | CCNA_01847 | 3.242669 | -0.248265 | 0.924899 | cytochrome c oxidase assembly protein Surf1 | energy metabolism |
| 64 | CCNA_01850 | 3.103049 | -0.394538 | 0.843285 | quinol cytochrome oxidase polypeptide I | energy metabolism |
| 63 | CCNA_01848 | 2.894681 | -0.344636 | 0.862939 | quinol cytochrome oxidase polypeptide IV | energy metabolism |
| 73 | CCNA_02723 | 2.486727 | 0.049502 | 0.923812 | PAS domain protein | signal transduction |
| 3 | CCNA_03933 | 2.474392 | 0.001013 | 0.999616 | hypothetical protein | hypothetical protein |
| 2 | CCNA_03932 | 2.331872 | -0.295854 | 0.946246 | hypothetical protein | hypothetical protein |
| 61 | CCNA_03770 | 2.313741 | 0.150902 | 0.502711 | malate dehydrogenase | Carbon metabolism |
| 22 | CCNA_R0117 | 1.857451 | -0.196825 | 0.467961 | small non-coding RNA | non-coding RNA |
| 72 | CCNA_02255 | 1.784785 | 0.040099 | 0.889995 | lysozyme-family localization factor spmX | miscellaneous |
| 18 | CCNA_00719 | 1.771677 | 0.181642 | 0.963599 | type I secretion adaptor protein hlyD | Transport |
| 0 | CCNA_00039 | 1.738843 | 0.496818 | 0.534773 | hypothetical protein | hypotetical protein |
| 16 | CCNA_00720 | 1.536174 | -0.059334 | 0.986843 | type I protein secretion ATP-binding protein | Transport |
| 78 | CCNA_02185 | -1.430931 | -0.367111 | 0.146861 | acetolactate synthase large subunit | Aminoacid metabolism |
| 23 | CCNA_02250 | -1.719603 | -0.110464 | 0.728662 | methylcrotonyl-CoA carboxylase biotin-containing subunit | Aminoacid metabolism |
| 79 | CCNA_03322 | -1.729614 | -0.175264 | 0.747879 | D-3-phosphoglycerate dehydrogenase | Aminoacid metabolism |
| 40 | CCNA_01521 | -1.733974 | 0.061491 | 0.837015 | hypothetical protein | hypothetical protein |
| 60 | CCNA_03248 | -1.806752 | -0.222469 | 0.564068 | TonB-dependent receptor | Transport |
| 57 | CCNA_00796 | -1.886055 | -0.203463 | 0.391589 | oxygen sensory histidine kinase FixL | Signal transduction |
| 24 | CCNA_02251 | -1.948257 | -0.099145 | 0.725245 | methylglutaconyl-CoA hydratase | Aminoacid metabolism |
6. Provenance and summary¶
In [8]:
prov = f'''# NB01 source provenance — Leaden 2018 Fur-only signature
Date: {pd.Timestamp.now().isoformat(timespec='seconds')}Z
Source: Leaden et al. 2018, Frontiers in Microbiology (PMID 30210482; PMC6120978)
File: ~/data/kr-caulobacter-envelope/raw/Table 2.XLSX (published supplementary)
Status: PREFLIGHT SUCCESS — CTS re-analysis from raw SRA was not required.
Sheets parsed: {[s for s in wb.sheetnames if not s.lower().startswith("summary")]}
Leaden Δfur DEGs (deduplicated): {len(leaden_fur)}
Leaden iron-limitation (2h DP) DEGs: {len(leaden_iron)}
'''
(DATA_OUT / 'NB01_source_provenance.md').write_text(prov)
print(prov)
# NB01 source provenance — Leaden 2018 Fur-only signature Date: 2026-06-04T02:22:07Z Source: Leaden et al. 2018, Frontiers in Microbiology (PMID 30210482; PMC6120978) File: ~/data/kr-caulobacter-envelope/raw/Table 2.XLSX (published supplementary) Status: PREFLIGHT SUCCESS — CTS re-analysis from raw SRA was not required. Sheets parsed: ['up wt-DP and fur ', 'down wt-DP and fur', 'up fur ', 'down fur', 'up wt-DP ', 'down wt-DP'] Leaden Δfur DEGs (deduplicated): 93 Leaden iron-limitation (2h DP) DEGs: 491
In [9]:
print('=== NB01 SUMMARY ===\n')
print(f'Leaden Δfur DEGs: {len(leaden_fur)}')
print(f'Leaden iron-limitation DEGs: {len(leaden_iron)}')
print(f'Shared (both Δfur AND iron): {len(leaden_merged.dropna(subset=["log2fc_fur","log2fc_iron_2hDP"]))}')
print()
print(f'Joined with our 4584-vs-4580: {len(both)} of {n_total} Leaden Δfur DEGs')
print(f'Spearman concordance: ρ = {rho_s:.3f}, p = {p_s:.2e}')
print(f'Sign concordance: {sign_concord:.1%}')
print(f'Stop condition (ρ ≥ 0.30): {VERDICT}')
print()
print('Category breakdown:')
print(cat_counts.to_string())
print()
print('NEXT — feed concordant_strong (core Fur signature) to NB02 for phenotype ranking via RB-TnSeq.')
=== NB01 SUMMARY === Leaden Δfur DEGs: 93 Leaden iron-limitation DEGs: 491 Shared (both Δfur AND iron): 62 Joined with our 4584-vs-4580: 93 of 93 Leaden Δfur DEGs Spearman concordance: ρ = 0.315, p = 2.08e-03 Sign concordance: 71.0% Stop condition (ρ ≥ 0.30): PASS — Δfur is a major driver Category breakdown: category_call buffered 53 concordant_strong 32 concordant_weak 5 discordant 3 NEXT — feed concordant_strong (core Fur signature) to NB02 for phenotype ranking via RB-TnSeq.