Tissue Expression Filter Test with TCGA LAML Data¶

Testing tissue expression functionality with real TCGA Acute Myeloid Leukemia data.

In [1]:

import sys
import os
import pandas as pd
from IPython.display import display

# Add src to path
sys.path.insert(0, os.path.join('..', '..', '..', 'src'))

from pyMut.input import read_maf
from pyMut.filters.tissue_expression import tissue_expression

import sys import os import pandas as pd from IPython.display import display # Add src to path sys.path.insert(0, os.path.join('..', '..', '..', 'src')) from pyMut.input import read_maf from pyMut.filters.tissue_expression import tissue_expression

Load TCGA LAML Dataset¶

In [2]:

# Load real TCGA LAML data
maf_path = os.path.join('..', '..', '..', 'src', 'pyMut', 'data', 'examples', 'MAF','tcga_laml.maf.gz')
# TCGA data is typically based on GRCh37 assembly
py_mut = read_maf(maf_path, assembly="37")

print(f"Loaded TCGA LAML data: {len(py_mut.data)} variants")
print(f"Unique genes: {py_mut.data['Hugo_Symbol'].nunique()}")
print(f"Unique samples: {py_mut.data['Tumor_Sample_Barcode'].nunique()}")

# Load real TCGA LAML data maf_path = os.path.join('..', '..', '..', 'src', 'pyMut', 'data', 'examples', 'MAF','tcga_laml.maf.gz') # TCGA data is typically based on GRCh37 assembly py_mut = read_maf(maf_path, assembly="37") print(f"Loaded TCGA LAML data: {len(py_mut.data)} variants") print(f"Unique genes: {py_mut.data['Hugo_Symbol'].nunique()}") print(f"Unique samples: {py_mut.data['Tumor_Sample_Barcode'].nunique()}")

2025-08-01 02:03:35,885 | INFO | pyMut.input | Starting MAF reading: ../../../src/pyMut/data/examples/MAF/tcga_laml.maf.gz
2025-08-01 02:03:35,886 | INFO | pyMut.input | Loading from cache: ../../../src/pyMut/data/examples/MAF/.pymut_cache/tcga_laml.maf_8bfbda65c4b23428.parquet
2025-08-01 02:03:35,911 | INFO | pyMut.input | Cache loaded successfully in 0.03 seconds

Loaded TCGA LAML data: 2091 variants
Unique genes: 1611
Unique samples: 190

Test 1: Individual Gene Expression (with prints)¶

In [3]:

# Test individual gene expression with genes from the dataset
sample_genes = py_mut.data['Hugo_Symbol'].value_counts().head(3).index.tolist()
print(f"Testing genes: {sample_genes}")

for gene in sample_genes:
    result_laml = tissue_expression(gene, ["LAML", 5])  # LAML = Acute Myeloid Leukemia
    result_blca = tissue_expression(gene, ["BLCA", 5])  # BLCA = Bladder Cancer
    print(f"{gene} - LAML (>5): {result_laml}, BLCA (>5): {result_blca}")

# Test individual gene expression with genes from the dataset sample_genes = py_mut.data['Hugo_Symbol'].value_counts().head(3).index.tolist() print(f"Testing genes: {sample_genes}") for gene in sample_genes: result_laml = tissue_expression(gene, ["LAML", 5]) # LAML = Acute Myeloid Leukemia result_blca = tissue_expression(gene, ["BLCA", 5]) # BLCA = Bladder Cancer print(f"{gene} - LAML (>5): {result_laml}, BLCA (>5): {result_blca}")

Testing genes: ['FLT3', 'DNMT3A', 'TET2']
FLT3 - LAML (>5): False, BLCA (>5): False
DNMT3A - LAML (>5): False, BLCA (>5): False
TET2 - LAML (>5): False, BLCA (>5): False

Test 2: PyMutation Object Filtering¶

In [4]:

print(f"Original TCGA data: {len(py_mut.data)} variants")

# Filter by LAML tissue (should keep many since this is LAML data)
filtered_laml = py_mut.filter_by_tissue_expression([('LAML', 5)])
print(f"LAML expressed (>5): {len(filtered_laml.data)} variants")


# Filter by multiple tissues
filtered_multi = py_mut.filter_by_tissue_expression([('LAML', 5), ('BRCA', 3), ('LUAD', 4)])
print(f"Multi-tissue expressed: {len(filtered_multi.data)} variants")


# Filter for NOT expressed in LAML with high threshold
filtered_not = py_mut.filter_by_tissue_expression([('LAML', 50)], keep_expressed=False)
print(f"NOT highly expressed in LAML: {len(filtered_not.data)} variants")

print(f"Original TCGA data: {len(py_mut.data)} variants") # Filter by LAML tissue (should keep many since this is LAML data) filtered_laml = py_mut.filter_by_tissue_expression([('LAML', 5)]) print(f"LAML expressed (>5): {len(filtered_laml.data)} variants") # Filter by multiple tissues filtered_multi = py_mut.filter_by_tissue_expression([('LAML', 5), ('BRCA', 3), ('LUAD', 4)]) print(f"Multi-tissue expressed: {len(filtered_multi.data)} variants") # Filter for NOT expressed in LAML with high threshold filtered_not = py_mut.filter_by_tissue_expression([('LAML', 50)], keep_expressed=False) print(f"NOT highly expressed in LAML: {len(filtered_not.data)} variants")

Original TCGA data: 2091 variants
LAML expressed (>5): 0 variants
Multi-tissue expressed: 915 variants
NOT highly expressed in LAML: 2091 variants

Test 3: Tissue Expression Results Analysis¶

In [5]:

# Display tissue expression results dataframes
print("📊 Detailed tissue expression analysis results:")

if hasattr(filtered_multi, 'tissue_expression_results'):
    results_df = filtered_multi.tissue_expression_results
    print(f"\nResults dataframe shape: {results_df.shape}")
    print("Columns:", list(results_df.columns))

    print("\n📊 Complete tissue expression results:")
    display(results_df)

    # Create summary statistics table
    summary_stats = []
    for col in results_df.columns:
        if col.endswith('_expressed'):
            tissue = col.replace('_expressed', '')
            count = results_df[col].sum()
            total = len(results_df)
            percentage = (count / total * 100) if total > 0 else 0
            summary_stats.append({
                'Tissue': tissue,
                'Expressed_Count': count,
                'Total_Genes': total,
                'Percentage': f"{percentage:.1f}%"
            })

    if summary_stats:
        print("\n📊 Summary statistics by tissue:")
        summary_df = pd.DataFrame(summary_stats)
        display(summary_df)
else:
    print("❌ Results dataframe not found")

# Display tissue expression results dataframes print("📊 Detailed tissue expression analysis results:") if hasattr(filtered_multi, 'tissue_expression_results'): results_df = filtered_multi.tissue_expression_results print(f"\nResults dataframe shape: {results_df.shape}") print("Columns:", list(results_df.columns)) print("\n📊 Complete tissue expression results:") display(results_df) # Create summary statistics table summary_stats = [] for col in results_df.columns: if col.endswith('_expressed'): tissue = col.replace('_expressed', '') count = results_df[col].sum() total = len(results_df) percentage = (count / total * 100) if total > 0 else 0 summary_stats.append({ 'Tissue': tissue, 'Expressed_Count': count, 'Total_Genes': total, 'Percentage': f"{percentage:.1f}%" }) if summary_stats: print("\n📊 Summary statistics by tissue:") summary_df = pd.DataFrame(summary_stats) display(summary_df) else: print("❌ Results dataframe not found")

📊 Detailed tissue expression analysis results:

Results dataframe shape: (2091, 9)
Columns: ['Index', 'Gene_Symbol', 'Expressed_in_Any_Tissue', 'LAML_expressed', 'LAML_threshold', 'BRCA_expressed', 'BRCA_threshold', 'LUAD_expressed', 'LUAD_threshold']

📊 Complete tissue expression results:

	Index	Gene_Symbol	Expressed_in_Any_Tissue	LAML_expressed	LAML_threshold	BRCA_expressed	BRCA_threshold	LUAD_expressed	LUAD_threshold
0	0	KIAA1529	False	False	5	False	3	False	4
1	1	KIAA1529	False	False	5	False	3	False	4
2	2	TBC1D2	True	False	5	True	3	True	4
3	3	LPPR1	False	False	5	False	3	False	4
4	4	BAAT	False	False	5	False	3	False	4
...	...	...	...	...	...	...	...	...	...
2086	2086	FRAS1	False	False	5	False	3	False	4
2087	2087	GPR78	False	False	5	False	3	False	4
2088	2088	MEPE	False	False	5	False	3	False	4
2089	2089	RAP1GDS1	True	False	5	True	3	True	4
2090	2090	SLC2A9	False	False	5	False	3	False	4

2091 rows × 9 columns

📊 Summary statistics by tissue:

	Tissue	Expressed_Count	Total_Genes	Percentage
0	LAML	0	2091	0.0%
1	BRCA	886	2091	42.4%
2	LUAD	754	2091	36.1%