cwas 0.1.2

Category-wide association study (CWAS). This is a data analytic tool to perform stringent association tests to find non-coding loci associated with autism spectrum disorder (ASD).

Category-wide association study (CWAS)

CWAS (Category-Wide Association Study) is a data analytic tool to perform stringent association tests to find non-coding loci associated with autism spectrum disorder (ASD). CWAS runs category-based burden tests using de novo variants from whole genome sequencing data and diverse annotation data sets.

CWAS was used in the following papers.

• An analytical framework for whole genome sequence association studies and its implications for autism spectrum disorder (Werling et al., 2018)
• Genome-wide de novo risk score implicates promoter variation in autism spectrum disorder (An et al., 2018)
• CWAS-Plus: Estimating genome-wide evaluation of noncoding variation from whole genome sequencing data. (Kim et al., in preperation)

Here is the original CWAS repository: sanderslab/cwas

Quickstart

Data requirements

Users must prepare following data for CWAS because it is very essential but cannot be generated automatically. Here are details.

1. Input VCF data (De novo variant list)

#CHROM  POS ID  REF ALT QUAL    FILTER  INFO
chr1    3747728 .        T       C       .       .       SAMPLE=11000.p1;BATCH=P231
chr1    38338861        .       C       A       .       .       SAMPLE=11000.p1;BATCH=P231
chr1    117942118       .      T       G       .       .       SAMPLE=11000.p1;BATCH=P231

• The input VCF data must follow the specification of VCF.
• The INFO field must contain a sample ID of each variant with this format SAMPLE={sample_id}.

2. List of samples

SAMPLE	FAMILY	PHENOTYPE
11000.p1	11000	case
11000.s1	11000	ctrl
11002.p1	11002	case
11002.s1	11002	ctrl

• CWAS requires the file like above listing sample IDs with its family IDs and phenotypes (Case=case, Control=ctrl).
• Here are details of the required format.
  • Tab separated
  • 3 essential columns: SAMPLE, FAMILY, and PHENOTYPE
  • A value in the PHENOTYPE must be case or ctrl.
• The values in the SAMPLE must be matched with the sample IDs of variants in the input VCF file.

3. List of adjustment factors (Optional)

SAMPLE	AdjustFactor
11000.p1	0.932
11000.s1	1.082
11002.p1	0.895
11002.s1	1.113

• The file like above is required if you want to adjust the number of variants for each sample in CWAS.
• Here are details of the required format.
  • Tab separated
  • 2 essential columns: SAMPLE and AdjustFactor
  • A value in the AdjustFactor must be a float.
• The values in the SAMPLE must be matched with the sample IDs of variants in the input VCF file.

You can get the examples of the above data requirements from joonan-lab/cwas-input-example

4. CWAS annotation files

You can install those file from this repository: joonan-lab/cwas-dataset

git clone https://github.com/joonan-lab/cwas-dataset.git

Due to the sizes of BigWig files for conservation scores, you must install them manually. Please follow this instruction.

Installation

We recomment using conda virtual environment to build environment for CWAS. Installing CWAS-Plus within a conda environment will prevent its installation in the global environment. When creating a conda environment, also install Python to enable local installations using pip. Run the following statements in your shell.

pip

conda env create -n cwas python=3.9
conda activate cwas
pip install cwas_plus

github

conda env create -n cwas python=3.9
conda activate cwas
git clone https://github.com/joonan-lab/cwas.git
cd cwas
pip install .

In addition, you must install Variant Effect Predictor (VEP).

CWAS Execution

1. Start

Run this command.

cwas start

As default, this command creates CWAS workspace in your home directory. The path is $HOME/.cwas. $HOME/.cwas/configuration.txt has also generated.

Alternatively, CWAS workspace can be specified with -w option. The configuration.txt will also be generated in the specified CWAS workspace.

.cwas
└── configuration.txt

2. Configuration

Write the following information in the $HOME/.cwas/configuration.txt.

ANNOTATION_DATA_DIR=/path/to/your/dir
GENE_MATRIX=/path/to/your/file
ANNOTATION_KEY_CONFIG=/path/to/your/file
VEP=/path/to/your/vep
VEP_CACHE_DIR=/path/to/your/vep/cache/dir
VEP_CONSERVATION_FILE=/path/to/your/vep/resource
VEP_LOFTEE=/path/to/your/vep/resource
VEP_HUMAN_ANCESTOR_FA=/path/to/your/vep/resource
VEP_GERP_BIGWIG=/path/to/your/vep/resource
VEP_MIS_DB=/path/to/your/missense/database
VEP_MIS_INFO_KEY=
VEP_MIS_THRES=

The ANNOTATION_DATA is a directory that contains all the BED files from joonan-lab/cwas-dataset. The VEP_MIS_DB is a database that is used to define damaging missense variants. The VEP_MIS_INFO_KEY is an user-defined name of the database used to annotate variants. The VEP_MIS_THRES is a threshold for missense variants (missense variants with value>=threshold are defined as damaging missense variants).

After writing the above file, run this command.

cwas configuration

Following files will be generated in your home directory as default. If you specify CWAS workspace, the files will be located in the same directory as the configuration.txt.

.cwas
├── annotation-data
├── annotation_keys.yaml
├── category_domain.yaml
├── configuration.txt
├── gene_matrix.txt
└── redundant_category.txt
.cwas_env

3. Preparation

This step merges the BED files to annotate variants. Run the following command.

cwas preparation -p 4

4 is the number of worker processes. You can adjust this.

After running this, merged BED file and its index will be generated in your CWAS workspace.

.cwas
...
├── merged_annotation.bed.gz
├── merged_annotation.bed.gz.tbi
...

4. Annotation

This step annotate your VCF file using VEP. Run this command.

cwas annotation -v /path/to/your/vcf -p 4

4 is the number of worker processes. You can adjust this.

Here is the result file.

.cwas
...
├── {Your VCF filename}.annotated.vcf.gz
...

5. Categorization

This step categorize your variants using the annotation datasets. Run this command.

cwas categorization -i /path/to/your/annotated/vcf -p 4

4 is the number of worker processes. You can adjust this.

After running this, you will get...

.cwas
...
├── {Your VCF filename}.categorization_result.txt.gz
...

6. Burden Test (Binomial Test)

This step is for calculation of relative risks and p-values for each category. As a default, these tests are based on variant-level analysis. The --use_n_carrier option can be used for sample-level analysis.

Binomial test

This step runs category-based burden test using the categorization result. The type of the test is binomial test. Run this command.

cwas binomial_test -i /path/to/your/categorization/result -s /path/to/your/samples [-a /path/to/your/adj_factors]

[] means that this is optional. If -a option is not specified, this step will bypass the adjustment step.

After running this, you will get...

.cwas
...
├── {Your VCF filename}.burden_test.txt
├── {Your VCF filename}.burden_test.volcano_plot.pdf
├── {Your VCF filename}.category_counts.txt
├── {Your VCF filename}.category_info.txt
...

Permutation test

This step runs category-based permutations using the categorization result. Run this command.

cwas permutation_test -i /path/to/your/categorization/result -s /path/to/your/samples [-a /path/to/your/adj_factors] [-b]

If -b option is specified, this step will generate binomial p-values for each permutation. This p-values will be used for burden shift and DAWN analysis.

After running this, you will get...

.cwas
...
├── {Your VCF filename}.permutation_test.txt
├── {Your VCF filename}.binom_pvals.txt.gz
...