### Install tensorQTL using pip

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Install tensorQTL using pip for easy setup.

```bash
pip3 install tensorqtl
```

--------------------------------

### Install Pgenlib from source

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Install Pgenlib from its source code if direct pip installation is not suitable.

```bash
git clone git@github.com:chrchang/plink-ng.git
cd plink-ng/2.0/Python/
python3 setup.py build_ext
python3 setup.py install
```

--------------------------------

### Install tensorQTL from repository with Conda

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Clone the repository and install tensorQTL using a Conda environment file.

```bash
$ git clone git@github.com:broadinstitute/tensorqtl.git
$ cd tensorqtl
# install into a new virtual environment and load
$ mamba env create -f install/tensorqtl_env.yml
$ conda activate tensorqtl
```

--------------------------------

### Import Libraries and Setup

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Imports necessary libraries for data manipulation, plotting, and TensorQTL. Sets up the computation device (GPU or CPU) and printing configurations.

```python
import pandas as pd
import numpy as np
import tensorqtl
from tensorqtl import pgen, cis, post
import qtl.plot

import matplotlib.pyplot as plt
import matplotlib
matplotlib.rcParams['figure.dpi'] = 72
plt.rcParams.update({'font.family': 'Helvetica', 'svg.fonttype':'none', 'pdf.fonttype':42})

import torch
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
print(f"torch: {torch.__version__} (CUDA {torch.version.cuda}), device: {device}")
print(f"pandas {pd.__version__}")
```

--------------------------------

### Install R and qvalue Package

Source: https://github.com/broadinstitute/tensorqtl/blob/master/install/INSTALL.md

Installs R and the 'qvalue' package using BiocManager, which is required for computing q-values.

```r
if (!require("BiocManager", quietly = TRUE))
    install.packages("BiocManager")
BiocManager::install("qvalue")
```

--------------------------------

### Setup and Data Loading for tensorQTL

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Initializes the environment, loads necessary libraries, and prepares genotype, phenotype, and covariate data for tensorQTL analysis. Ensure data paths are correctly set.

```python
import pandas as pd
import torch
import tensorqtl
from tensorqtl import pgen, cis, trans, post
device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
print(f"torch: {torch.__version__} (CUDA {torch.version.cuda}), device: {device}")
print(f"pandas: {pd.__version__}")

# define paths to data
plink_prefix_path = 'data/GEUVADIS.445_samples.GRCh38.20170504.maf01.filtered.nodup.chr18'
expression_bed = 'data/GEUVADIS.445_samples.expression.bed.gz'
covariates_file = 'data/GEUVADIS.445_samples.covariates.txt'
prefix = 'GEUVADIS.445_samples'

# load phenotypes and covariates
phenotype_df, phenotype_pos_df = tensorqtl.read_phenotype_bed(expression_bed)
covariates_df = pd.read_csv(covariates_file, sep='\t', index_col=0).T

# PLINK reader for genotypes
pgr = pgen.PgenReader(plink_prefix_path)
genotype_df = pgr.load_genotypes()
variant_df = pgr.variant_df
```

--------------------------------

### Install latest tensorQTL from Git repository

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Install the latest version of tensorQTL directly from its GitHub repository using pip.

```bash
pip install pip@git+https://github.com/broadinstitute/tensorqtl.git
```

--------------------------------

### Install CUDA Drivers

Source: https://github.com/broadinstitute/tensorqtl/blob/master/install/INSTALL.md

Installs CUDA drivers on Ubuntu 22.04 LTS. Requires a reboot to take effect. Verify installation with nvidia-smi.

```bash
sudo ./install_cuda.sh
sudo reboot
# verify
nvidia-smi
```

--------------------------------

### Install Pgenlib for PLINK 2

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Install the Pgenlib Python package, required for using PLINK 2 binary files.

```bash
pip install Pgenlib
```

--------------------------------

### Install rmate (Optional)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/install/INSTALL.md

Installs the rmate tool, which allows remote editing of files. It requires Ruby and adds the rmate executable to the user's bin directory and configures the bashrc.

```bash
sudo apt install -y ruby
mkdir ~/bin
curl -Lo ~/bin/rmate https://raw.githubusercontent.com/textmate/rmate/master/bin/rmate
chmod a+x ~/bin/rmate
echo 'export RMATE_PORT=${rmate_port}' >> ~/.bashrc
```

--------------------------------

### Verify PyTorch and CUDA Installation

Source: https://github.com/broadinstitute/tensorqtl/blob/master/install/INSTALL.md

Verifies the installed PyTorch version and checks if CUDA is available, printing the CUDA device name if available. Expected output format is shown.

```python
import torch; print(torch.__version__); print('CUDA available: {} ({})'.format(torch.cuda.is_available(), torch.cuda.get_device_name(torch.cuda.current_device())))
```

--------------------------------

### Run tensorQTL help command

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Display all available command-line options for tensorQTL.

```bash
python3 -m tensorqtl --help
```

--------------------------------

### Create and Activate Conda Environment

Source: https://github.com/broadinstitute/tensorqtl/blob/master/install/INSTALL.md

Creates a conda environment using the provided 'tensorqtl_env.yml' file and activates it. This environment includes PyTorch and TensorQTL.

```bash
mamba env create -f tensorqtl_env.yml
conda activate tensorqtl
```

--------------------------------

### Run cis-QTL mapping with permutations (Shell)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Execute this shell command to perform cis-QTL mapping. The `${prefix}` argument specifies the output file name.

```shell
python3 -m tensorqtl ${plink_prefix_path} ${expression_bed} ${prefix} \
    --covariates ${covariates_file} \
    --mode cis
```

--------------------------------

### Run trans-QTL mapping (Shell)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Execute this shell command to perform trans-QTL mapping. By default, it generates sparse output for associations with p-value < 1e-5 and filters out cis-associations.

```shell
python3 -m tensorqtl ${plink_prefix_path} ${expression_bed} ${prefix} \
    --covariates ${covariates_file} \
    --mode trans
```

--------------------------------

### Map cis-QTL interactions (Shell)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Use this shell command to map cis-QTLs with an interaction term. The `--interaction` option specifies the interaction file. `--best_only` disables full summary statistics output.

```shell
python3 -m tensorqtl ${plink_prefix_path} ${expression_bed} ${prefix} \
    --covariates ${covariates_file} \
    --interaction ${interactions_file} \
    --best_only \
    --mode cis_nominal
```

--------------------------------

### Run trans-QTL mapping (Python)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

This Python code performs trans-QTL mapping, generating sparse output of associations with p-value < 1e-5. It includes filtering of cis-associations.

```python
trans_df = trans.map_trans(genotype_df, phenotype_df, covariates_df,
                           return_sparse=True, pval_threshold=1e-5, maf_threshold=0.05,
                           batch_size=20000)
# remove cis-associations
trans_df = trans.filter_cis(trans_df, phenotype_pos_df.T.to_dict(), variant_df, window=5000000)
```

--------------------------------

### Run cis-QTL mapping with permutations (Python)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Use this Python code to perform cis-QTL mapping and generate phenotype-level summary statistics with empirical p-values for genome-wide FDR calculation.

```python
cis_df = cis.map_cis(genotype_df, variant_df, phenotype_df, phenotype_pos_df, covariates_df)
tensorqtl.calculate_qvalues(cis_df, qvalue_lambda=0.85)
```

--------------------------------

### Import tensorQTL modules

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Import necessary pandas and tensorQTL modules for running analyses in Python.

```python
import pandas as pd
import tensorqtl
from tensorqtl import genotypeio, cis, trans
```

--------------------------------

### Run cis-QTL Mapping and Calculate Q-values

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Performs cis-QTL mapping using the loaded genotype, phenotype, and covariate data. It then calculates q-values based on the empirical p-values with a specified FDR.

```python
# run permutations and compute q-values
cis_df = cis.map_cis(genotype_df, pgr.variant_df, phenotype_df, phenotype_pos_df,
                     covariates_df=covariates_df, seed=3042235018, 
                     warn_monomorphic=False)
post.calculate_qvalues(cis_df, fdr=0.05, qvalue_lambda=0.85)
```

--------------------------------

### Load Genotypes

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Loads genotype data from a PLINK prefix file using PgenReader. This is a prerequisite for running the cis-QTL analysis.

```python
plink_prefix_path = '/resources/vcfs/GTEx_Analysis_2017-06-05_v8_WholeGenomeSeq_838Indiv_Analysis_Freeze.SHAPEIT2_phased.MAF01'
pgr = pgen.PgenReader(plink_prefix_path)
genotype_df = pgr.load_genotypes()
```

--------------------------------

### Generate cis-QTL summary statistics (Shell)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Use this shell command to generate nominal cis-QTL associations. The results are saved to parquet files per chromosome.

```shell
python3 -m tensorqtl ${plink_prefix_path} ${expression_bed} ${prefix} \
    --covariates ${covariates_file} \
    --mode cis_nominal
```

--------------------------------

### Calculate Q-values for Cis-Associations

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Calculates empirical q-values from a DataFrame of cis-associations. Requires a DataFrame with cis-associations and specifies FDR and q-value lambda parameters.

```python
post.calculate_qvalues(cis_df, fdr=0.05, qvalue_lambda=0.85)
```

--------------------------------

### Read cis-QTL summary statistics from Parquet (Python)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

This Python snippet demonstrates how to read the parquet files containing cis-QTL summary statistics using pandas.

```python
df = pd.read_parquet(file_name)
```

--------------------------------

### Generate cis-QTL summary statistics (Python)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

This Python code generates nominal associations for all variant-phenotype pairs. Results are written to parquet files.

```python
cis.map_nominal(genotype_df, variant_df, phenotype_df, phenotype_pos_df,
                prefix, covariates_df, output_dir='.')
```

--------------------------------

### Map conditionally independent cis-QTLs (Shell)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Execute this shell command to map conditionally independent cis-QTLs. The `--cis_output` option specifies the input from the permutation step.

```shell
python3 -m tensorqtl ${plink_prefix_path} ${expression_bed} ${prefix} \
    --covariates ${covariates_file} \
    --cis_output ${prefix}.cis_qtl.txt.gz \
    --mode cis_independent
```

--------------------------------

### Plotting q-value comparison with zoom

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

This snippet visualizes the comparison of q-values between FastQTL v8 and cis-QTL results, with a zoomed-in view on a specific threshold.

```python
# zoom on FDR threshold
ax[3].set_xscale('log')
ax[3].set_yscale('log')
ax[3].scatter(egenes_v8_df['qval'], cis_df['qval'], **args)
ax[3].set_title('q-values (zoom)', fontsize=12)
ax[3].set_xlabel('FastQTL (v8)', fontsize=12)
b = [0.01, 0.1]
ax[3].plot(b, [0.05, 0.05], 'k--')
ax[3].plot([0.05, 0.05], b, 'k--')
ax[3].set_xlim(b)
ax[3].set_ylim(b)
ax[3].set_xticks([0.01, 0.05, 0.1])
ax[3].set_yticks([0.01, 0.05, 0.1]);
```

--------------------------------

### Load subset of samples with PlinkReader

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Load genotypes using PlinkReader, selecting only a specific subset of samples to potentially save memory.

```python
pr = genotypeio.PlinkReader(plink_prefix_path, select_samples=phenotype_df.columns)
```

--------------------------------

### Perform Trans-QTL Mapping

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Initiates trans-QTL mapping using genotype, phenotype, and covariate data. Allows for batch processing, sparse output, and filtering by p-value and minor allele frequency.

```python
# run mapping
# to limit output size, only associations with p-value <= 1e-5 are returned
trans_df = trans.map_trans(genotype_df, phenotype_df, covariates_df, batch_size=10000,
                           return_sparse=True, pval_threshold=1e-5, maf_threshold=0.05)
```

--------------------------------

### Load genotypes from PLINK reader

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Initialize a PlinkReader and load genotypes and variant information into pandas DataFrames.

```python
pr = genotypeio.PlinkReader(plink_prefix_path)
# load genotypes and variants into data frames
genotype_df = pr.load_genotypes()
variant_df = pr.bim.set_index('snp')[['chrom', 'pos']]
```

--------------------------------

### Run Nominal cis-QTL Mapping

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Performs nominal cis-QTL mapping for a specified chromosome, saving the results to a specified output directory. Requires genotype, phenotype, and covariate data.

```python
# nominal mode (limit to one chromosome for illustration purposes)
chrom = 'chr9'
m = phenotype_pos_df['chr'] == chrom
cis.map_nominal(genotype_df, pgr.variant_df, phenotype_df[m], phenotype_pos_df[m], tissue_id,
                covariates_df=covariates_df, output_dir='/mnt/disks/scratch/tmp/')
nominal_df = pd.read_parquet(f'/mnt/disks/scratch/tmp/Brain_Cortex.cis_qtl_pairs.{chrom}.parquet')
```

--------------------------------

### Map cis-QTL interactions with eigenMT (Python)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

This Python code maps cis-QTLs including an interaction term and computes eigenMT-adjusted p-values. It requires interaction data and can write top associations and full summary statistics.

```python
cis.map_nominal(genotype_df, variant_df, phenotype_df, phenotype_pos_df, prefix,
                covariates_df=covariates_df,
                interaction_df=interaction_df, maf_threshold_interaction=0.05,
                run_eigenmt=True, output_dir='.', write_top=True, write_stats=True)
```

--------------------------------

### Load GTEx v8 Results and Plot Comparisons

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Loads pre-computed GTEx v8 eQTL results and generates plots to compare nominal p-values, empirical p-values, and q-values between the GTEx v8 results and the TensorQTL output.

```python
# load V8 results and plot comparisons
egenes_v8_df = pd.read_csv(f'/resources/V8_release/GTEx_Analysis_v8_eQTL/{tissue_id}.v8.egenes.txt.gz',
                           sep='\t', index_col=0)

args = {'ec':'none', 'alpha':0.5}
ax = qtl.plot.get_axgrid(1, 4)
ax[0].set_xscale('log')
ax[0].set_yscale('log')
ax[0].scatter(egenes_v8_df['pval_nominal'], cis_df['pval_nominal'], **args)
ax[0].set_title('Nominal p-values', fontsize=12)
ax[0].set_xlabel('FastQTL (v8)', fontsize=12)
ax[0].set_ylabel('TensorQTL', fontsize=12)

ax[1].set_xscale('log')
ax[1].set_yscale('log')
ax[1].scatter(egenes_v8_df['pval_beta'], cis_df['pval_beta'], **args)
ax[1].set_title('Empirical p-values', fontsize=12)
ax[1].set_xlabel('FastQTL (v8)', fontsize=12)

ax[2].set_xscale('log')
ax[2].set_yscale('log')
ax[2].scatter(egenes_v8_df['qval'], cis_df['qval'], **args)
ax[2].set_title('q-values', fontsize=12)
ax[2].set_xlabel('FastQTL (v8)', fontsize=12);
```

--------------------------------

### Load phenotypes and covariates

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Load phenotype data from a BED file and covariates from a tab-delimited file into pandas DataFrames.

```python
phenotype_df, phenotype_pos_df = tensorqtl.read_phenotype_bed(phenotype_bed_file)
covariates_df = pd.read_csv(covariates_file, sep='\t', index_col=0).T  # samples x covariates
```

--------------------------------

### Load and Inspect Cis-QTL Nominal Results

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Loads the results of the cis-QTL nominal association mapping from a Parquet file and displays the first few rows. This is useful for reviewing the calculated p-values and effect sizes.

```python
# load results
pairs_df = pd.read_parquet(f'{prefix}.cis_qtl_pairs.chr18.parquet')
pairs_df.head()
```

--------------------------------

### Map conditionally independent cis-QTLs (Python)

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

This Python code maps conditionally independent cis-QTLs using a stepwise regression procedure. It requires the output from the permutation step.

```python
indep_df = cis.map_independent(genotype_df, variant_df, cis_df,
                               phenotype_df, phenotype_pos_df, covariates_df)
```

--------------------------------

### Load and Subset V8 Summary Statistics

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Loads GTEx v8 summary statistics for a given tissue and subsets them to a specific chromosome. Resets the index after subsetting.

```python
# load V8 summary statistics
nominal_v8_df = pd.read_parquet(
    f"gs://gtex-resources/GTEx_Analysis_v8_QTLs/GTEx_Analysis_v8_eQTL_all_associations/{tissue_id}.v8.allpairs.parquet")

# subset chr9
nominal_v8_df = nominal_v8_df[nominal_v8_df['variant_id'].str.startswith(chrom)]
nominal_v8_df.reset_index(drop=True, inplace=True)
```

--------------------------------

### Calculate eGene Overlap

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Calculates and prints the number of eGenes and their intersection between FastQTL v8 and TensorQTL results at a 0.05 q-value threshold.

```python
# eGenes overlap
igenes_v8 = egenes_v8_df[egenes_v8_df['qval'] <= 0.05].index
egenes = cis_df[cis_df['qval'] <= 0.05].index
ix = egenes[egenes.isin(egenes_v8)]
print(f"eGenes, v8 (all biotypes):   {len(egenes_v8)}")
print(f"eGenes, rep. (all biotypes): {len(egenes)}")
print(f"  * intersection:            {len(ix)} ({len(ix)/len(egenes_v8)*100:.1f}%)")
```

--------------------------------

### Load Phenotype and Covariate Data

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Loads normalized expression data (phenotypes) and covariate data for a specific tissue. Phenotype data is loaded from a BED file, and covariates from a tab-separated text file.

```python
tissue_id = 'Brain_Cortex'
covariates_df = pd.read_csv(f"/resources/V8_release/GTEx_Analysis_v8_eQTL_covariates/{tissue_id}.v8.covariates.txt",
                            sep='\t', index_col=0).T
phenotype_df, phenotype_pos_df = tensorqtl.read_phenotype_bed(
    f'/resources/V8_release/GTEx_Analysis_v8_eQTL_expression_matrices/{tissue_id}.v8.normalized_expression.bed.gz')
```

--------------------------------

### Display Head of Cis-Association DataFrame

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Displays the first few rows of a pandas DataFrame containing cis-association results. Useful for inspecting the structure and content of the data.

```python
cis_df.head()
```

--------------------------------

### Convert VCF to PLINK2 pgen format

Source: https://github.com/broadinstitute/tensorqtl/blob/master/README.md

Convert VCF files to PLINK2 pgen/pvar/psam format using plink2. Ensure to use --output-chr chrM for compatibility.

```bash
plink2 \
    --output-chr chrM \
    --vcf ${plink_prefix_path}.vcf.gz \
    --out ${plink_prefix_path}
```

--------------------------------

### Plot Nominal p-value Comparison

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Generates a scatter plot comparing nominal p-values from GTEx v8 and TensorQTL results. Both axes are set to a logarithmic scale.

```python
# plot p-values
ax = qtl.plot.get_axgrid(1, 1)[0]
ax.set_xscale('log')
ax.set_yscale('log')
ax.scatter(nominal_v8_df['pval_nominal'], nominal_df['pval_nominal'], **args)
ax.set_title('Nominal p-values', fontsize=12)
ax.set_xlabel('FastQTL p-value', fontsize=12)
ax.set_ylabel('TensorQTL p-value', fontsize=12);
```

--------------------------------

### Display Head of Trans-Association DataFrame

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Displays the first few rows of a pandas DataFrame containing trans-association results after filtering. Useful for inspecting the filtered data.

```python
trans_df.head()
```

--------------------------------

### Compare V8 and TensorQTL Nominal p-values

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/GTEx_v8_example.ipynb

Asserts the equality of gene IDs, variant IDs, and TSS distances between GTEx v8 and TensorQTL nominal results for a specific chromosome. This is a validation step before plotting.

```python
# check that outputs are identical
assert nominal_v8_df['gene_id'].equals(nominal_df['phenotype_id'])
assert nominal_v8_df['variant_id'].equals(nominal_df['variant_id'])
assert nominal_v8_df['tss_distance'].equals(nominal_df['start_distance'])
```

--------------------------------

### Cis-QTL Empirical P-value Calculation

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Calculates empirical p-values for cis-QTL associations using a specified random seed for reproducibility. This snippet also focuses on a specific chromosome (chr18).

```python
# all genes
# cis_df = cis.map_cis(genotype_df, variant_df, phenotype_df, phenotype_pos_df, covariates_df=covariates_df)

# genes on chr18
cis_df = cis.map_cis(genotype_df, variant_df, 
                     phenotype_df.loc[phenotype_pos_df['chr'] == 'chr18'],
                     phenotype_pos_df.loc[phenotype_pos_df['chr'] == 'chr18'],
                     covariates_df=covariates_df, seed=123456)
```

--------------------------------

### Cis-QTL Nominal Association Mapping

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Performs cis-QTL mapping to calculate nominal p-values for all variant-phenotype pairs within a specified window. This snippet focuses on a specific chromosome (chr18).

```python
# map all cis-associations (results for each chromosome are written to file)

# all genes
# cis.map_nominal(genotype_df, variant_df, phenotype_df, phenotype_pos_df, prefix, covariates_df=covariates_df)

# genes on chr18
cis.map_nominal(genotype_df, variant_df,
                phenotype_df.loc[phenotype_pos_df['chr'] == 'chr18'],
                phenotype_pos_df.loc[phenotype_pos_df['chr'] == 'chr18'],
                prefix, covariates_df=covariates_df)
```

--------------------------------

### Filter Cis-Associations from Trans-QTL Results

Source: https://github.com/broadinstitute/tensorqtl/blob/master/example/tensorqtl_examples.ipynb

Removes cis-associations from a trans-QTL results DataFrame. Requires the trans-QTL DataFrame, phenotype position data, variant data, and a window size for filtering.

```python
# remove cis-associations
trans_df = trans.filter_cis(trans_df, phenotype_pos_df, variant_df, window=5000000)
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.