04 Module Analysis
Jupyter notebook from the Field vs Lab Gene Importance in Desulfovibrio vulgaris Hildenborough project.
NB 04: Module-Level Analysis¶
Test whether ICA fitness modules responding to field-relevant conditions are more conserved (higher core gene fraction) than modules responding to lab-only conditions.
Analyses:
- Module conservation score (fraction of member genes that are core)
- Module ecological relevance (mean activity across field vs lab conditions)
- Correlation:
module_core_fraction ~ module_field_activity - Module characterization: ecological vs lab modules
- Gene-level candidates from ecological modules
Run locally — uses cached data from upstream projects + NB02 output.
Outputs: 2 figures in figures/, module characterization table.
In [1]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
from scipy import stats
from pathlib import Path
plt.rcParams['figure.dpi'] = 150
plt.rcParams['savefig.dpi'] = 150
plt.rcParams['savefig.bbox'] = 'tight'
DATA_DIR = Path('../data')
FIG_DIR = Path('../figures')
FIG_DIR.mkdir(exist_ok=True)
1. Load Data¶
In [2]:
# Module membership: gene x module binary matrix
membership = pd.read_csv(
'../../fitness_modules/data/modules/DvH_gene_membership.csv',
index_col=0
)
print(f"Module membership: {membership.shape[0]} genes x {membership.shape[1]} modules")
# Module-condition activity scores
mod_cond = pd.read_csv(
'../../fitness_modules/data/modules/DvH_module_conditions.csv'
)
print(f"Module-condition activity: {len(mod_cond)} rows")
print(f" Modules: {mod_cond['module'].nunique()}")
print(f" Experiments: {mod_cond['expName'].nunique()}")
# Module annotations (functional enrichments)
mod_annot = pd.read_csv(
'../../fitness_modules/data/modules/DvH_module_annotations.csv'
)
print(f"Module annotations: {len(mod_annot)} rows")
# Experiment classification from NB02
exp_class = pd.read_csv(DATA_DIR / 'experiment_classification.csv')
print(f"Experiment classification: {len(exp_class)} experiments")
# Pangenome link (DvH only)
link = pd.read_csv(
'../../conservation_vs_fitness/data/fb_pangenome_link.tsv',
sep='\t'
)
link_dvh = link[link['orgId'] == 'DvH'].copy()
link_dvh['locusId'] = link_dvh['locusId'].astype(int)
print(f"Pangenome link (DvH): {len(link_dvh)} genes")
# SEED annotations
seed = pd.read_csv(
'../../conservation_vs_fitness/data/seed_annotations.tsv',
sep='\t', dtype={'gene': str}
)
seed_dvh = seed[seed['orgId'] == 'DvH'].copy()
print(f"SEED annotations (DvH): {len(seed_dvh)} genes")
Module membership: 2741 genes x 52 modules Module-condition activity: 260 rows Modules: 52 Experiments: 198 Module annotations: 140 rows Experiment classification: 757 experiments
Pangenome link (DvH): 3206 genes SEED annotations (DvH): 2815 genes
2. Module Conservation Score¶
For each of 52 modules, compute fraction of member genes that are core.
In [3]:
# Build gene -> is_core mapping
# Handle string 'True'/'False' or actual bools
is_core_raw = link_dvh.set_index('locusId')['is_core']
if is_core_raw.dtype == object:
gene_core = is_core_raw.map({'True': True, 'False': False}).to_dict()
else:
gene_core = is_core_raw.fillna(False).astype(bool).to_dict()
# For each module, compute conservation
module_stats = []
modules = membership.columns.tolist()
for mod in modules:
member_genes = membership.index[membership[mod] == 1].tolist()
n_members = len(member_genes)
if n_members == 0:
continue
# Count core genes
n_core = sum(1 for g in member_genes if gene_core.get(g, False))
n_with_link = sum(1 for g in member_genes if g in gene_core)
module_stats.append({
'module': mod,
'n_members': n_members,
'n_with_link': n_with_link,
'n_core': n_core,
'core_fraction': n_core / n_with_link if n_with_link > 0 else np.nan
})
mod_stats = pd.DataFrame(module_stats)
print(f"Module conservation scores: {len(mod_stats)} modules")
print(f" Mean core fraction: {mod_stats['core_fraction'].mean():.3f}")
print(f" Median core fraction: {mod_stats['core_fraction'].median():.3f}")
print(f" Range: {mod_stats['core_fraction'].min():.3f} - {mod_stats['core_fraction'].max():.3f}")
mod_stats.head(10)
Module conservation scores: 52 modules Mean core fraction: 0.886 Median core fraction: 1.000 Range: 0.000 - 1.000
Out[3]:
| module | n_members | n_with_link | n_core | core_fraction | |
|---|---|---|---|---|---|
| 0 | M000 | 8 | 8 | 6 | 0.750000 |
| 1 | M001 | 3 | 3 | 3 | 1.000000 |
| 2 | M002 | 19 | 19 | 19 | 1.000000 |
| 3 | M003 | 9 | 9 | 7 | 0.777778 |
| 4 | M004 | 9 | 9 | 8 | 0.888889 |
| 5 | M005 | 12 | 12 | 10 | 0.833333 |
| 6 | M006 | 15 | 15 | 14 | 0.933333 |
| 7 | M007 | 17 | 17 | 8 | 0.470588 |
| 8 | M008 | 7 | 7 | 7 | 1.000000 |
| 9 | M009 | 4 | 4 | 4 | 1.000000 |
3. Module Ecological Relevance¶
For each module, compute mean absolute activity across field vs lab conditions.
In [4]:
# Map experiments to broad categories
exp_to_cat = exp_class.set_index('expName')['category'].to_dict()
exp_to_broad = exp_class.set_index('expName')['broad_category'].to_dict()
# Add category to module-condition table
mod_cond['category'] = mod_cond['expName'].map(exp_to_cat)
mod_cond['broad_category'] = mod_cond['expName'].map(exp_to_broad)
print("Module-condition activity by broad category:")
print(mod_cond.groupby('broad_category')['abs_activity'].agg(['mean', 'count']))
Module-condition activity by broad category:
mean count
broad_category
field 0.240946 131
lab 0.231699 129
In [5]:
# Compute mean absolute activity per module per broad category
mod_activity = mod_cond.groupby(['module', 'broad_category'])['abs_activity'].mean().unstack(fill_value=0)
mod_activity.columns = [f'mean_activity_{c}' for c in mod_activity.columns]
mod_activity = mod_activity.reset_index()
# Also compute per fine category
mod_activity_fine = mod_cond.groupby(['module', 'category'])['abs_activity'].mean().unstack(fill_value=0)
mod_activity_fine = mod_activity_fine.reset_index()
print(f"Module activity (broad): {mod_activity.shape}")
print(f"Module activity (fine): {mod_activity_fine.shape}")
# Merge with conservation stats
mod_combined = mod_stats.merge(mod_activity, on='module', how='left')
mod_combined = mod_combined.merge(mod_activity_fine, on='module', how='left', suffixes=('', '_fine'))
# Compute field/lab activity ratio
if 'mean_activity_field' in mod_combined.columns and 'mean_activity_lab' in mod_combined.columns:
mod_combined['field_lab_ratio'] = (
mod_combined['mean_activity_field'] /
mod_combined['mean_activity_lab'].replace(0, np.nan)
)
print(f"\nCombined module table: {len(mod_combined)} modules")
mod_combined[['module', 'n_members', 'core_fraction',
'mean_activity_field', 'mean_activity_lab']].head(10)
Module activity (broad): (52, 3) Module activity (fine): (52, 7) Combined module table: 52 modules
Out[5]:
| module | n_members | core_fraction | mean_activity_field | mean_activity_lab | |
|---|---|---|---|---|---|
| 0 | M000 | 8 | 0.750000 | 0.208600 | 0.000000 |
| 1 | M001 | 3 | 1.000000 | 0.208901 | 0.000000 |
| 2 | M002 | 19 | 1.000000 | 0.000000 | 0.150004 |
| 3 | M003 | 9 | 0.777778 | 0.139379 | 0.123832 |
| 4 | M004 | 9 | 0.888889 | 0.000000 | 0.290945 |
| 5 | M005 | 12 | 0.833333 | 0.000000 | 0.299652 |
| 6 | M006 | 15 | 0.933333 | 0.000000 | 0.321461 |
| 7 | M007 | 17 | 0.470588 | 0.000000 | 0.165578 |
| 8 | M008 | 7 | 1.000000 | 0.288439 | 0.143942 |
| 9 | M009 | 4 | 1.000000 | 0.000000 | 0.179172 |
4. Correlation: Module Conservation vs Field Activity¶
In [6]:
# Test correlation
valid = mod_combined.dropna(subset=['core_fraction', 'mean_activity_field'])
if len(valid) > 5:
r_field, p_field = stats.spearmanr(valid['core_fraction'], valid['mean_activity_field'])
r_lab, p_lab = stats.spearmanr(valid['core_fraction'], valid['mean_activity_lab'])
print("Spearman correlation (module core fraction vs activity):")
print(f" Field activity: rho={r_field:.3f}, p={p_field:.4g}")
print(f" Lab activity: rho={r_lab:.3f}, p={p_lab:.4g}")
if 'field_lab_ratio' in valid.columns:
valid_ratio = valid.dropna(subset=['field_lab_ratio'])
r_ratio, p_ratio = stats.spearmanr(valid_ratio['core_fraction'],
valid_ratio['field_lab_ratio'])
print(f" Field/Lab ratio: rho={r_ratio:.3f}, p={p_ratio:.4g}")
else:
print("Not enough modules for correlation analysis")
Spearman correlation (module core fraction vs activity): Field activity: rho=0.071, p=0.6152 Lab activity: rho=0.041, p=0.7719 Field/Lab ratio: rho=-0.139, p=0.4203
In [7]:
# Figure 1: Scatter of module core fraction vs field condition activity
fig, axes = plt.subplots(1, 2, figsize=(14, 6))
# Panel A: Field activity vs conservation
ax = axes[0]
ax.scatter(valid['mean_activity_field'], valid['core_fraction'],
s=valid['n_members'] * 3, alpha=0.6, edgecolor='black', linewidth=0.5,
c='#2ca02c')
# Add labels for extreme modules
for _, row in valid.nlargest(3, 'mean_activity_field').iterrows():
ax.annotate(row['module'], (row['mean_activity_field'], row['core_fraction']),
fontsize=7, ha='left')
for _, row in valid.nsmallest(3, 'core_fraction').iterrows():
ax.annotate(row['module'], (row['mean_activity_field'], row['core_fraction']),
fontsize=7, ha='left')
ax.set_xlabel('Mean abs. activity (field conditions)')
ax.set_ylabel('Core genome fraction')
ax.set_title(f'Module conservation vs field activity\n'
f'(Spearman rho={r_field:.3f}, p={p_field:.3g})')
ax.set_ylim(-0.05, 1.05)
# Panel B: Lab activity vs conservation
ax = axes[1]
ax.scatter(valid['mean_activity_lab'], valid['core_fraction'],
s=valid['n_members'] * 3, alpha=0.6, edgecolor='black', linewidth=0.5,
c='#d62728')
for _, row in valid.nlargest(3, 'mean_activity_lab').iterrows():
ax.annotate(row['module'], (row['mean_activity_lab'], row['core_fraction']),
fontsize=7, ha='left')
for _, row in valid.nsmallest(3, 'core_fraction').iterrows():
ax.annotate(row['module'], (row['mean_activity_lab'], row['core_fraction']),
fontsize=7, ha='left')
ax.set_xlabel('Mean abs. activity (lab conditions)')
ax.set_ylabel('Core genome fraction')
ax.set_title(f'Module conservation vs lab activity\n'
f'(Spearman rho={r_lab:.3f}, p={p_lab:.3g})')
ax.set_ylim(-0.05, 1.05)
plt.tight_layout()
plt.savefig(FIG_DIR / 'fig_module_conservation_vs_activity.png')
plt.show()
5. Ecological vs Lab Module Characterization¶
Classify modules as "ecological" (field-active + highly conserved) vs "lab" (lab-active + less conserved).
In [8]:
# Classify modules
# Use mean instead of median for core_fraction threshold, because
# median=1.0 would prevent any module from being classified as "ecological"
median_field = mod_combined['mean_activity_field'].median()
mean_core = mod_combined['core_fraction'].mean()
print(f"Classification thresholds:")
print(f" Field activity > median = {median_field:.4f}")
print(f" Core fraction > mean = {mean_core:.3f} (median was {mod_combined['core_fraction'].median():.3f})")
print()
def classify_module(row):
high_field = row['mean_activity_field'] > median_field
high_core = row['core_fraction'] > mean_core
if high_field and high_core:
return 'ecological'
elif not high_field and not high_core:
return 'lab'
elif high_field and not high_core:
return 'field-variable'
else:
return 'conserved-quiet'
mod_combined['module_type'] = mod_combined.apply(classify_module, axis=1)
print("Module classification:")
print(mod_combined['module_type'].value_counts())
Classification thresholds: Field activity > median = 0.1731 Core fraction > mean = 0.886 (median was 1.000) Module classification: module_type ecological 21 conserved-quiet 17 lab 9 field-variable 5 Name: count, dtype: int64
In [9]:
# Add functional annotations for each module
# Get top annotation per module (lowest FDR)
sig_annot = mod_annot[mod_annot['significant'] == True].copy()
if len(sig_annot) > 0:
top_annot = sig_annot.sort_values('fdr').groupby('module').first().reset_index()
top_annot = top_annot[['module', 'term', 'database', 'odds_ratio', 'fdr']]
top_annot.columns = ['module', 'top_term', 'top_db', 'top_or', 'top_fdr']
mod_combined = mod_combined.merge(top_annot, on='module', how='left')
# Show ecological modules
print("Ecological modules (field-active + conserved):")
eco_mods = mod_combined[mod_combined['module_type'] == 'ecological']
display_cols = ['module', 'n_members', 'core_fraction',
'mean_activity_field', 'mean_activity_lab']
if 'top_term' in mod_combined.columns:
display_cols += ['top_term', 'top_db']
eco_mods[display_cols].sort_values('mean_activity_field', ascending=False)
Ecological modules (field-active + conserved):
Out[9]:
| module | n_members | core_fraction | mean_activity_field | mean_activity_lab | top_term | top_db | |
|---|---|---|---|---|---|---|---|
| 23 | M023 | 11 | 1.000000 | 0.370467 | 0.000000 | Phosphonate ABC transporter phosphate-binding ... | SEED |
| 22 | M022 | 21 | 0.952381 | 0.337329 | 0.000000 | PF04879 | PFam |
| 30 | M030 | 9 | 0.888889 | 0.326659 | 0.000000 | FIG00603085: hypothetical protein | SEED |
| 51 | M051 | 16 | 1.000000 | 0.306403 | 0.000000 | PF13533 | PFam |
| 28 | M028 | 11 | 0.909091 | 0.296810 | 0.000000 | PF00037 | PFam |
| 8 | M008 | 7 | 1.000000 | 0.288439 | 0.143942 | PF12679 | PFam |
| 32 | M032 | 7 | 1.000000 | 0.286228 | 0.000000 | K10040 | KEGG |
| 43 | M043 | 3 | 1.000000 | 0.280256 | 0.000000 | NaN | NaN |
| 35 | M035 | 37 | 0.944444 | 0.270945 | 0.000000 | FIG00603085: hypothetical protein | SEED |
| 40 | M040 | 4 | 1.000000 | 0.260326 | 0.000000 | K00532 | KEGG |
| 42 | M042 | 5 | 1.000000 | 0.252021 | 0.000000 | NaN | NaN |
| 48 | M048 | 3 | 1.000000 | 0.251500 | 0.136722 | PF02589 | PFam |
| 45 | M045 | 4 | 1.000000 | 0.231140 | 0.270388 | NaN | NaN |
| 29 | M029 | 7 | 1.000000 | 0.225582 | 0.302252 | CBS domain protein | SEED |
| 1 | M001 | 3 | 1.000000 | 0.208901 | 0.000000 | NaN | NaN |
| 19 | M019 | 20 | 0.894737 | 0.204631 | 0.193789 | PF00148 | PFam |
| 11 | M011 | 50 | 1.000000 | 0.201332 | 0.121071 | PF00460 | PFam |
| 49 | M049 | 8 | 1.000000 | 0.197130 | 0.000000 | NaN | NaN |
| 14 | M014 | 1 | 1.000000 | 0.194955 | 0.272925 | NaN | NaN |
| 46 | M046 | 6 | 1.000000 | 0.191635 | 0.282134 | NaN | NaN |
| 27 | M027 | 6 | 1.000000 | 0.183191 | 0.174607 | NaN | NaN |
In [10]:
# Show lab modules
print("Lab modules (lab-biased + less conserved):")
lab_mods = mod_combined[mod_combined['module_type'] == 'lab']
lab_mods[display_cols].sort_values('mean_activity_lab', ascending=False)
Lab modules (lab-biased + less conserved):
Out[10]:
| module | n_members | core_fraction | mean_activity_field | mean_activity_lab | top_term | top_db | |
|---|---|---|---|---|---|---|---|
| 5 | M005 | 12 | 0.833333 | 0.000000 | 0.299652 | NaN | NaN |
| 34 | M034 | 15 | 0.866667 | 0.000000 | 0.289211 | PF00528 | PFam |
| 44 | M044 | 6 | 0.333333 | 0.124774 | 0.283917 | PF13641 | PFam |
| 7 | M007 | 17 | 0.470588 | 0.000000 | 0.165578 | PF13186 | PFam |
| 20 | M020 | 41 | 0.685714 | 0.153807 | 0.150508 | K02003 | KEGG |
| 12 | M012 | 17 | 0.000000 | 0.000000 | 0.132460 | TIGR01539 | TIGRFam |
| 3 | M003 | 9 | 0.777778 | 0.139379 | 0.123832 | PF00072 | PFam |
| 31 | M031 | 36 | 0.676471 | 0.135532 | 0.113038 | NaN | NaN |
| 17 | M017 | 6 | 0.000000 | 0.118333 | 0.107751 | NaN | NaN |
In [11]:
# Figure 2: Comparison of ecological vs lab module profiles
fig, axes = plt.subplots(1, 2, figsize=(12, 5))
# Panel A: Core fraction by module type
ax = axes[0]
type_order = ['ecological', 'conserved-quiet', 'field-variable', 'lab']
type_colors = {'ecological': '#2ca02c', 'conserved-quiet': '#1f77b4',
'field-variable': '#ff7f0e', 'lab': '#d62728'}
plot_types = [t for t in type_order if t in mod_combined['module_type'].values]
box_data = [mod_combined[mod_combined['module_type'] == t]['core_fraction'].values
for t in plot_types]
bp = ax.boxplot(box_data, labels=plot_types, patch_artist=True)
for patch, t in zip(bp['boxes'], plot_types):
patch.set_facecolor(type_colors.get(t, '#7f7f7f'))
patch.set_alpha(0.7)
ax.set_ylabel('Core genome fraction')
ax.set_title('Module conservation by type')
ax.set_xticklabels(plot_types, rotation=20, ha='right')
# Panel B: Field vs lab activity colored by module type
ax = axes[1]
for t in type_order:
subset = mod_combined[mod_combined['module_type'] == t]
if len(subset) == 0:
continue
ax.scatter(subset['mean_activity_field'], subset['mean_activity_lab'],
c=type_colors.get(t, '#7f7f7f'), label=f'{t} (n={len(subset)})',
s=subset['n_members'] * 3, alpha=0.7, edgecolor='black', linewidth=0.5)
# Diagonal
lim = max(ax.get_xlim()[1], ax.get_ylim()[1])
ax.plot([0, lim], [0, lim], 'k--', linewidth=0.5, alpha=0.3)
ax.set_xlabel('Mean abs. activity (field)')
ax.set_ylabel('Mean abs. activity (lab)')
ax.set_title('Module activity: field vs lab')
ax.legend(fontsize=8, loc='upper left')
plt.tight_layout()
plt.savefig(FIG_DIR / 'fig_ecological_vs_lab_modules.png')
plt.show()
/tmp/ipykernel_31185/1059578287.py:14: MatplotlibDeprecationWarning: The 'labels' parameter of boxplot() has been renamed 'tick_labels' since Matplotlib 3.9; support for the old name will be dropped in 3.11. bp = ax.boxplot(box_data, labels=plot_types, patch_artist=True)
6. Gene-Level Candidates from Ecological Modules¶
Genes in ecological modules that lack annotation are candidates for novel environmental adaptation functions.
In [12]:
# Get genes from ecological modules
eco_module_names = eco_mods['module'].tolist()
eco_genes = []
for mod in eco_module_names:
if mod not in membership.columns:
continue
member_genes = membership.index[membership[mod] == 1].tolist()
for g in member_genes:
eco_genes.append({'locusId': int(g), 'module': mod})
eco_gene_df = pd.DataFrame(eco_genes)
print(f"Genes in ecological modules: {len(eco_gene_df)}")
if len(eco_gene_df) > 0:
# Merge with SEED annotations
seed_cols = seed_dvh.columns.tolist()
desc_col = 'seed_desc' if 'seed_desc' in seed_cols else 'description' if 'description' in seed_cols else None
if desc_col and 'locusId' in seed_cols:
seed_dvh_subset = seed_dvh[['locusId', desc_col]].copy()
seed_dvh_subset['locusId'] = seed_dvh_subset['locusId'].astype(int)
eco_gene_df = eco_gene_df.merge(seed_dvh_subset, on='locusId', how='left')
# Merge with conservation
core_map = link_dvh[['locusId', 'is_core']].copy()
if core_map['is_core'].dtype == object:
core_map['is_core'] = core_map['is_core'].map({'True': True, 'False': False}).astype(bool)
else:
core_map['is_core'] = core_map['is_core'].fillna(False).astype(bool)
eco_gene_df = eco_gene_df.merge(core_map, on='locusId', how='left')
# Identify unannotated genes
if desc_col and desc_col in eco_gene_df.columns:
unannotated = eco_gene_df[
eco_gene_df[desc_col].isna() |
eco_gene_df[desc_col].str.contains('hypothetical|uncharacterized', case=False, na=True)
]
print(f"Unannotated genes in ecological modules: {len(unannotated)}")
print(f" Of which core: {unannotated['is_core'].sum()}")
print()
print("These are candidates for novel environmental adaptation functions.")
unannotated.head(20)
else:
print("SEED annotation column not found; showing all ecological genes.")
eco_gene_df.head(20)
else:
print("No ecological modules found (none had both high field activity and high conservation).")
print("This may indicate that conservation and field activity are not strongly correlated at the module level.")
unannotated = pd.DataFrame()
Genes in ecological modules: 239 Unannotated genes in ecological modules: 52 Of which core: 44 These are candidates for novel environmental adaptation functions.
7. Summary¶
In [13]:
print("=" * 60)
print("MODULE-LEVEL ANALYSIS SUMMARY")
print("=" * 60)
print()
print(f"Modules analyzed: {len(mod_combined)}")
print(f" Mean core fraction: {mod_combined['core_fraction'].mean():.3f}")
print(f" Median core fraction: {mod_combined['core_fraction'].median():.3f}")
print()
print("Module classification:")
for t in ['ecological', 'conserved-quiet', 'field-variable', 'lab']:
n = (mod_combined['module_type'] == t).sum()
if n > 0:
cf = mod_combined[mod_combined['module_type'] == t]['core_fraction'].mean()
print(f" {t:20s}: {n:3d} modules, mean core fraction = {cf:.3f}")
print()
print("Correlations (Spearman):")
try:
print(f" Core fraction vs field activity: rho={r_field:.3f}, p={p_field:.4g}")
print(f" Core fraction vs lab activity: rho={r_lab:.3f}, p={p_lab:.4g}")
except NameError:
print(" (correlation variables not computed)")
print()
print(f"Genes in ecological modules: {len(eco_gene_df)}")
if 'unannotated' in dir() and len(unannotated) > 0:
print(f" Unannotated candidates: {len(unannotated)}")
print("=" * 60)
============================================================ MODULE-LEVEL ANALYSIS SUMMARY ============================================================ Modules analyzed: 52 Mean core fraction: 0.886 Median core fraction: 1.000 Module classification: ecological : 21 modules, mean core fraction = 0.980 conserved-quiet : 17 modules, mean core fraction = 0.983 field-variable : 5 modules, mean core fraction = 0.829 lab : 9 modules, mean core fraction = 0.516 Correlations (Spearman): Core fraction vs field activity: rho=0.071, p=0.6152 Core fraction vs lab activity: rho=0.041, p=0.7719 Genes in ecological modules: 239 Unannotated candidates: 52 ============================================================
In [14]:
# Save module characterization table
mod_combined.to_csv(DATA_DIR / 'module_characterization.csv', index=False)
print(f"Saved: {DATA_DIR / 'module_characterization.csv'}")
Saved: ../data/module_characterization.csv