| title | date | output | ||||||
|---|---|---|---|---|---|---|---|---|
metaGWASmanager documentation |
13/12/2023 |
|
This vignette introduces the metaGWASmanager, a comprehensive toolbox leveraging existing software packages and, streamlining the entire GWAS-consortium workflow. This encompasses the phenotype generation, quality control of phenotypes, GWAS, GWAS-QC and meta-analysis providing an integrated solution for both the participating Study Analysts (SA) and Consortium Analysts (CA). For a description, please see the corresponding manuscript: Zulema Rodriguez-Hernandez, Mathias Gorski, Maria Tellez Plaza, Pascal Schlosser* and Matthias Wuttke* (2024). “metaGWASmanager: A toolbox for an automated workflow from phenotypes to meta-analysis in GWAS consortia”. Bioinformatics,2024. 10.1093/bioinformatics/btae294.
Illustration of the metaGWASmanager pipeline, outlining phenotype generation and quality assurance, followed by the GWAS, GWAS-QC and meta-analysis steps, along with the required inputs and resulting outputs. Shaded and white sections indicate tasks to be carried out by CA and SA, respectively.
To set up the toolbox please fork the GitHub repository: https://github.com/genepi-freiburg/gwas-consortium/tree/main
Aside from the documentation and scripts supplied by metaGWASmanager, it is necessary to ensure that the following programs and files are properly installed and downloaded on your system.
- Association tool to use. (e.g., regenie)
- PLINK or PLINK 2.0
- VCFtools and BCFtools
- R software
- Analysis plan, scripts and examples files for conducting Phases 2 and 4 will be provided by the CAs.
Same as SAs, but also:
- GWASinspector R package
- SQLite reference data available here
- The alt_headers.txt file is available in the extended data in the CRAN Archive
- Perl
- HTSlib
- FlexiBLAS
- Python
We outline the recommended structure for the consortium server to ensure effective and efficient data management:
They should be separate from data, in a different folder.
"/storage/consortium_name/scripts/"
This folder will contain the scripts needed by CAs, all provided by metaGWASmanager. It is advisable to directly download the different folders-containing scripts (pheno_generation, pheno_summary, gwas_qc and metaanalysis) from metaGWASmanager GitHub to the scripts folder previously created on your server:
- pheno_generation folder: it contains scripts needed by SAs to prepare phenotype data, and perform GWAS. To be used by SAs.
- pheno_summary folder: it contains scripts to summarize phenotype submissions across studies. For CAs' use.
- gwas_qc folder: it contains scripts to summarize genotype submissions across studies. For CAs' use.
- metaanalysis folder: it contains scripts to prepare data and perform meta-analysis. For CAs' use.
Please, do not modify the scripts names.
The phenotypes results, output of Phase 2 - SA: Prepare Phenotypes (SAs), submitted by the SAs will be stored in the uploads/pheno directory. This directory will contain a folder for each study that has submitted the phenotype summary statistics.
"/storage/consortium_name/uploads/pheno/study_name"
Besides the study-specific folders, it should also contain two additional directories:
- 00_ARCHIVE. If a specific-study re-upload the phenotype summary statistics, perhaps due to the detection of an error, CAs will archive the previous version in this directory.
- 00_SUMMARY. The output from performing cross-study phenotype summaries, in the Phase 3 - CA: Check & Approve Phenotypes Files, will be stored in this location.
The GWAS results (output of Phase 4 - SA: Perform Associations Analysis) submitted by the SAs will be stored in the uploads/assoc directory. This directory will contain a folder for each study that has submitted genetic summary statistics.
"/storage/consortium_name/uploads/assoc/study_name"
Similar to the uploads/pheno directory, we recommend to create a 00_ARCHIVE sub-directory.
It is advised to maintain consistency in the naming of studies within both the uploads/pheno and uploads/assoc folders to prevent potential problems when cross-referencing files between the two folders.
Contains intermediate files, logs and summary files resulting from Phase 5 - CA: Check Associations & Meta-analysis.
"/storage/consortium_name/cleaning/study_name"
In addition to the folders designated for each study, it should also encompass two extra directories:
- 00_ARCHIVE. To save older results.
- 00_SUMMARY. It will contain the output of cross-study GWAS verification process (Phase 5 - CA: Check Associations & Meta-analysis) including plots, overall statistical summaries, positive control validation results, among others.
Note that the sub-folder "data" within each study folder contains the GWAS results combined by chromosomes.
"/storage/consortium_name/cleaning/study_name/data"
Finally, the meta-analysis folder will contain as many subfolders as target traits (e.g, uacr_int trait).
"/storage/consortium_name/metaanalysis/uacr_int"
The initial stage involves generating Consortium-specific files (consortium-specifics.R and parameters.txt plug-in) and formulating the Analysis Plan, in which CAs will define, based on the provided metaGWASmanager examples, the required settings for conducting the pool-cohort GWAS.
A) consortium-specifics.R. The consortium-specifics.R plug-in file (located in the pheno_generation folder) should be edited according to the CA specifications. This file contains several functions that allow customization of the entire analysis process: verification of the input and parameteres files; unit conversion; traits transformations (rank-based inverse normalization and log-transformation are implemented); setting quality control parameters, covariates and stratification-variables; association tool to be used (e.g., PLINK 2.0, regenie, etc.), among others. These functions will be applied in subsequent stages of the workflow process. The consortium-specifics.R file will be provided to each SAs along the Analysis Plan and required scripts.
Please, note that, regarding the type of association tool to use, currently metaGWASmanager has implemented PLINK 2.0 and regenie softwares. CAs can choose one of these two. However, in case CAs prefer to use another different one, we provide a plug-in interface, assoc_tool_plugin.R, within the pheno_generation/assoc_tool folder, that CAs can customize to add a different association tool.
B) Parameters file. CA will customize the parameters.txt file based on the particular traits to study. It must included important points, which must be filled out by the SAs, such as the name of the input file and the participant-study, SAs' contact information, the study's ancestral background, analysis date, units of the variables to be studied, laboratory particular settings (e.g., limits of detection), imputation reference panel, number of genetic principal components, and some more optional fields. A parameters file example is provided within the pheno_generation folder on the metaGWASmanager GitHub repository.
C) Analysis Plan. The Analysis Plan (e.g., PDF document) should provided comprehensive information for SAs on how to proceed with the different analyses, including Phase 2 - SA: Prepare Phenotypes and Phase 4 - SA: Perform Associations Analysis. Certainly, you are welcome to use the information described in this documentation. It is also essential to include CAs' contact details in case any questions arise.
In a first step, CA will ask for preparation the phenotypic data for all participating studies. CA will supply an Analysis Plan and the pheno_generation folder. Please note, that for each study and for each ancestry, one copy of the pheno_generation folder should be used.
Input files for each participating study include: a) the input.txt file, a database created by SA according to the CA guide standards (see the dummy-input.txt example), and b) SAs will complete the parameter file (parameters.txt) provided by CAs.
After that, SAs will execute the 01-phenotype-generation.sh script, which automatically generates the descriptive statistics of phenotypes, on a Linux server as follows:
bash 01-phenotype-generation.sh parameters.txt
Note that if SAs prepare their data in a Windows or Mac system, to avoid problems with line breaks when running the scripts in Linux/Unix, they should use the following commands, respectively, to convert the files prior to running the scripts:
dos2unix input.txt
mac2unix input.txt
During this initial phase, a comprehensive examination of the input and parameters files will be carried out. This involves issuing warnings and identifying errors to guarantee high-quality phenotypic data and prevent common pitfalls such as unit conversion discrepancies or variations in assay methods. Additionally, the pipeline executes essential trait transformations (e.g., rank-based inverse normalization, log-transformation) and calculations, ensuring uniformity in the phenotype data.
The files generated by the phenotype preparation scripts include the phenotype summary statistics (STUDY_NAME_ids_summary.txt) and plots files. SAs must carefully examine the .log files for errors or warnings, as well as ensure that the generated descriptive statistics and plots (PDF file) are reasonable. If there are problems, SAs will try to adjust the parameters in the parameters.txt or input.txt file according to the different error messages.
In case everything seems fine, SAs will send all the files from the generated return_pheno folder via email to CAs. The CAs will then check phenotype descriptive data. It is important to note that the files uploaded by SAs contain only summarized statistical (not individual-level information).
You can find an example of a phenotype generation report here. It includes descriptive graphs (such as box plots and histograms) for the trait-specific variables (genetic principal components, outcomes, covariates, and stratified variables). When applicable, an informative bar plot is provided for each binary and categorical variable.
The submited files within the return_pheno folder will be stored in the upload/pheno directory by CAs, within the folder corresponding to the specific participating study.
"/storage/consortium_name/uploads/pheno/study_name"
Subsequently, CA will manually inspect these files and plots in order to identify potential issues and inconsistencies, such as anormal distributions ando/or outliers. Furthermore, CA will run the scripts located in the pheno_summary directory of the metaGWASmanager toolkit (01_collect_summary.sh and 02_plot_summaries.sh).
Both scripts summarize phenotype summary-statistics submissions (STUDY_NAME_ids_summary.txt) across studies and also facilitate their representation for improved inter-study comparison and to detect potential outliers. Please run both scripts using the bash command line from pheno_summary folder:
"/storage/consortium_name/scripts/pheno_summary"
The generated outputs of the inter-study comparison analyses will be automatically saved in the 00_SUMMARY folder located at:
"/storage/consortium_name/uploads/pheno/00_SUMMARY"
After the inspection of the submitted files, CA will provide feedback to SAs. In the case CA detects an issue, it will be clearly outlined and assistance will be offered to the SAs in oder to resolve it. The SAs will then revisit Phase 2 - SA: Prepare Phenotypes and re-submit the corresponding outputs files. Contrary, if no problems are identified, SAs are authorized to proceed with Phase 4 - SA: Perform Associations Analysis.
To initiate this step, SAs should have received the approval from CA. First SAs prepared the genetic information according to the CAs guidelines.
For both association tools (PLINK 2.0 and regenie, specially for regenie-step2) imputed genetic data (chunked by chromosome) in .bgen format v1.2 with 8 bits encoding is required. SAs can easily convert the imputed data to .bgen format using the provided vcf_to_bgen.sh script, inside the pheno_generation folder or by running the following commands on Linux:
plink2 --vcf $infile dosage=DS --make-pgen erase-phase --out $pgen_unphased
plink2 --pfile $pgen_unphased --export bgen-1.2 bits=8 --out $out_bgen
For the regenie association tool, the measured genotype (chip data) in PLINK format is also required for the regenie-step1. We recommend to extract the high quality variants following the regenie best practices:
plink2 --geno 0.1 --hwe 1e-15 --mac 100 --maf 0.01 --mind 0.1 --indep-pairwise 1000 100 0.9
For that, SAs can use the provided plink_qc.sh script.
Once the genetic data is prepared, the SAs can systematically initiate the necessary script executions.
A) Create job files.
First, SAs should adjust the make-association-job-scripts within the pheno_generation/assoc_tool folder according to their requirements (data path and genotype file prefix, etc.).
A.1) For regenie:
Two files to be modified by SAs: make-regenie-step1-job-scripts.sh and make-regenie-step2-job-scripts.sh:
For make-regenie-step1-job-scripts.sh:
. REGENIE=<path to your regenie software>
. PLINK_DATA_PREFIX=<path and prefix to your PLINK *.bed file>
. PLINK_SNP_QC=<path to your *.snplist file containing a list of high quality variants used for regenie-step1>
. PLINK_INDIV_QC=<path to your *.id file containing a list of high-quality individuals used for regenie-step1>
. --threads <Number of threads that should be utilized by each individual regenie-step1 job>
For make-regenie-step2-job-scripts.sh:
. REGENIE=<path to your regenie software>
. BGEN_SAMPLE_PREFIX=<path to your *.bgen and *.sample files> (as one sample file exists per chromosome; please adjust the prefix file name using ${CHR} as placeholder for the chromosome).
. --threads <Number of threads that should be utilized by each individual regenie-step-2 job>
A.2) For PLINK 2.0:
One file to be edited by SAs: make-plink-job-scripts.sh:
. PLINK=<path to your PLINK 2.0 software>
. BGEN_SAMPLE_PREFIX=<path to your *.bgen and *.sample files> (as one sample file exists per chromosome; please adjust the prefix file name using ${CHR} as placeholder for the chromosome).
. --threads <Number of threads that should be utilized by each individual plink job>
Note that these scripts are set up to utilize a Slurm job scheduler.
Upon making the mentioned adjustments, SAs can execute the 02-consortium-make-assoc-jobs.sh script.
bash 02-consortium-make-assoc-jobs.sh parameters.txt
The process will generate the different jobs (saved in the jobs folder), which include the phenotypes and covariates to be used along with the command lines necessary to run the association analyses.
For instance, for regenie-step1, as many jobs as phenotype will be created, versus one job for each phenotype and chromosome for regenie-step2 and PLINK 2.0.
B) Run GWAS
SAs then will execute 03-submit-all-jobs.sh to submit all GWAS according to the jobs generated in the preceding step.
bash 03-submit-all-jobs.sh parameters.txt
Our recommendation is to first run a pilot analysis before submitting all jobs. For example, if regenie is used, one phenotype for regenie-step1 and one chromosome for regenie-step2.
C) Create Summary Statistics and Collect & Upload results.
The SAs will then run the 04-postprocess-results.sh script, which investigates the different log files generated during the association process to ensure the successful execution of each GWAS. It also produces tailored summary tables for each association analysis.
bash 04-postprocess-results.sh parameters.txt
If an issue occurs, SAs should investigate the cause. After resolving it, SAs can re-execute individual phenotypes and/or steps by using the appropriate files from the jobs directory. In case SAs needed to change any parameter o paths, they would have to re-run the 02-consortium-make-assoc-jobs.sh again.
Finally, the results will be compiled into a compressed folder using 05-collect-files-for-upload.sh
bash 05-collect-files-for-upload.sh study_name
A gzipped tar-ball folder will be created, containing outputs from return_pheno folder including log files, summary tables and plots, and the association-output folders (regenie-step2 or output_plink folder). The compressed folder will then be submitted to the CA server for further validation.
We also recommend that CAs, at this point, somehow collect information about the specific authors (names, affiliations, conflict of interest, etc.) and studies (genotyping information, study acknowledgements, other relevant study information). For example, CAs may use a Google Sheets and/or Google Forms.
The compressed folder submitted by the SAs, will be stored in the upload/assoc directory, within the folder corresponding to the specific participating study.
"/storage/consortium_name/uploads/assoc/study_name"
CA will then proceed to unzip the folder and run scripts located in the gwas_qc folder of the metaGWASmanager toolbox. Please, note that scripts located in gwas_qc will be run directly from this folder.
"/storage/consortium_name/scripts/gwas_qc"
It includes details about essential pathways (script folder, cleaning directory, upload data, etc.) and specific variables (phenotypes, ancestry and strata vectors, among others) needed for executing downstream scripts. CAs first should customize these fields and may also incorporate important settings to align with particular server requirements. The folders.config.sh file will be sourced throughout subsequent scripts, so no modification to other scripts are needed.
After making the necessary settings to the folders.config.sh file, CA are ready to carry out the subsequent steps.
A) 01_combine_chromosomes.sh.
It checks if any file is missing from the association-output folder ensuring the presence of the required columns using the find-column-index.pl script. Subsequently, it merges all the chromosomes into a single compressed file for each trait and the corresponding tabix files are generated.
Command line to run:
bash 01_combine_chromosomes.sh study_name
B) 02_gwasinspector.sh.
CAs will then run GWASinspector:
bash 02_gwasinspector.sh study_name
GWASinspector performs an extensive quality control process on GWAS results ensuring their accuracy, proper format, and consistency across all studies using robust criteria. As a result, in the cleaning/study_name/qc_output folder, clean and harmonized files will be created along with a comprehensive report and plots (including Manhattan, QQ plots, histograms, etc.) as shown below:
Manhattan plot created by GWASinspector workflow of overall eGFR creatinine in the GCKD cohort, a participating study in the CKDGenR5 consortium.
Quality controls plot created by GWASinspector workflow of overall eGFR creatinine in the GCKD cohort, a participating study in the CKDGenR5 consortium.
More information about GWASinspector R package here.
C) 03_check_positive_controls.sh
CAs should create a GWAS positive control .txt file containing the hits (specific to each target trait and ancestry) widely validated in the literature, whenever possible.
CAs can also find an example in the gwas_qc/Positive_Controls folder (please, replace the provided example by your own customized positive-controls.txt file):
| trait | ethnicity | SNP | gene | chr | pos_b37 | pos_b38 | ref | alt | MAF_1KGP | direction_alt |
|---|---|---|---|---|---|---|---|---|---|---|
| egfr_creat_int | EUR | rs13329952 | UMOD | 16 | 20366507 | 20355185 | T | C | 0.19 | pos |
| egfr_creat_male | EUR | rs13329952 | UMOD | 16 | 20366507 | 20355185 | T | C | 0.19 | pos |
| egfr_creat_female | EUR | rs13329952 | UMOD | 16 | 20366507 | 20355185 | T | C | 0.19 | pos |
| egfr_creat_int | EAS | rs10277115 | UNCX | 7 | 1285195 | 1245559 | A | T | 0.31 | neg |
| egfr_creat_male | EAS | rs10277115 | UNCX | 7 | 1285195 | 1245559 | A | T | 0.31 | neg |
| egfr_creat_female | EAS | rs10277115 | UNCX | 7 | 1285195 | 1245559 | A | T | 0.31 | neg |
By running 03_check_positive_controls.sh, each file within the cleaning directory will be checked and compared with the created positive-controls.txt
bash 03_check_positive_controls.sh study_name
Results will be saved in 03-check-positive-controls.csv file within the corresponding folder:
"/storage/consortium_name/cleaning/study_name"
D) 04_collect_qc_stats.sh
It checks all clean studies after GWASinspector quality control:
"/storage/consortium_name/cleaning/study_name/qc_output"
and generate a qc-stats.csv and qc-stats.xlsx files in the 00_SUMMARY folder:
"/storage/consortium_name/cleaning/00_SUMMARY"
The qc-stats.csv contains information about the specific study, phenotype, ancestry, sample size, variants per chromosome, lambda and summary stats for both "All" and "High quality" variants, such as P-value (PVAL), effect allele frequency (EFF_ALL_FREQ), imputation quality (IMP_QUALITY), effect size (BETA), standard error (STDERR). The file then, will contain as many rows as specific studies and phenotypes.
| STUDY | PHENO | POP | PVALUE_MED_ALL | EFF_ALL_FREQ_MED_ALL | IMP_QUALITY_MED_ALL | BETA_MED_ALL | STDERR_MED_ALL | STDERR_MED_HQ | INPUT_VARIANT_COUNT | HQ_VARIANT_COUNT | AF_CORRELATION_ALL | LAMBDA | SAMPLE_SIZE | VARIANTS_CHR_1 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GCKD_2022-10-04 | ckd | EUR | 0.4822 | 0.001215 | 0.9469 | 0.01283 | 0.7595 | 0.07497 | 28713489 | 9179280 | 0.957 | 1.09 | 4993 | 2,193,932 |
| GCKD_2022-10-04 | ma | EUR | 0.4787 | 0.001851 | 0.9572 | 0.008682 | 0.6586 | 0.08028 | 26066764 | 9180489 | 0.956 | 1.1 | 3998 | 1,985,785 |
| GCKD_2022-10-04 | gout | EUR | 0.4743 | 0.001198 | 0.9466 | -0.0164 | 0.7984 | 0.07839 | 28814280 | 9179205 | 0.957 | 1.13 | 5034 | 2,201,949 |
Command line to run:
bash 04_collect_qc_stats.sh
In addition, this script generates a general (all specific studies) positive controls summary files (positive-controls.csv and positive-controls.xlsx) in the 00_SUMMARY folder.
E) 05_plot_qc_stats.sh
Multi-studies summary plots will be then generated from qc-stats.csv file within the 00_SUMMARY/plots folder:
"/storage/consortium_name/cleaning/00_SUMMARY/plots"
Command line to run:
bash 05_plot_qc_stats.sh
F) 06_plot_frequencies.sh
Running 06_plot_frequencies.sh generates a correlation plot for each phenotype across all the specific studies. The objective is to verify the correlation of allele frequencies between the observed data with a reference panel.
bash 06_plot_frequencies.sh
Plots will be save in study-specific folder.
"/storage/consortium_name/cleaning/study_name/freqs"
Frequency correlation plot of overall eGFR creatinine in the GCKD cohort, a study participating in the CKDGenR5 consortium.
G) 07_stateOfAffairs.sh
Efficient monitoring of the consortium's progress, such as ongoing GWAS status, studies awaiting analysis, and current sample sizes is crucial for effective coordination and management of the consortium effort. The 07_stateOfAffairs.sh print such report and a pie plot from qc-stats.csv file.
Command line to run:
bash 07_stateOfAffairs.sh
H) 08_checkAllFileNames.sh
For proper execution of subsequent steps (Auto-GWAS quality control and meta-analysis) a final check is important to ensure consistency in files names, studies and phenotypes. The 08_checkAllFileNames.sh scripts systematically validate these aspects and report if any inconsistencies arise.
bash 08_checkAllFileNames.sh
I) 09_GWAS_QC_multistudies.sh
By combining the summary statistics from GWAS-QC (qc-stats.csv) and the results of positive controls analysis (positive-controls.csv) across all participating studies, CA will perform a final checking process. A set of relevant items will be checked, including: wrong genomic build, allele switches and swaps, missing data, file and number formatting issues, unaccounted inflation, and wrong transformations, among others.
MetaGWASmanager provides a script (09_GWAS_QC_multistudies.sh), which automatically evaluates the list of potential problems mentioned above. For its execution, CA will require the previously edited consortium-specifics.R plug-in, located in the pheno_generation folder. It supplies a tolerance table establishing the thresholds to apply at each verification stage based on the different phenotypes and covariates to be tested.
Command line to run:
bash 09_GWAS_QC_multistudies.sh
In the end, a conclusive-summary table will be generated in the cleaning/00_SUMMARY/AutoQC folder, indicating whether each trait for each assessed item is categorized as "OKAY" or "NOT OKAY".
| study | pheno | pop | N | pval | lambda | impqual | beta | stderr | allel_freq | vars.per.chr | positive_control |
|---|---|---|---|---|---|---|---|---|---|---|---|
| GCKD_2022-10-04 | ckd | EUR | 4993 | OK | OK | OK | OK | OK | OK | OK | OK |
| GCKD_2022-10-04 | ma | EUR | 3998 | OK | NOT OK | OK | OK | OK | OK | OK | OK |
| GCKD_2022-10-04 | gout | EUR | 5034 | OK | NOT OK | OK | OK | OK | OK | OK | OK |
| GCKD_2022-10-04 | albumin_serum_int | EUR | 4992 | OK | OK | OK | NOT OK | OK | OK | OK | OK |
| GCKD_2022-10-04 | egfr_creat_int | EUR | 4993 | OK | OK | OK | OK | OK | OK | OK | OK |
CA should scan all entries and emphasizing those labeled as "NOT OKAY", as well as, plots produced by GWASinspector step. If an explanation for any identified issues cannot be determined, CAs must reach out to the SAs to collaboratively work towards a resolution.
After completing all GWAS quality controls checks, CAs are pepared to proceed with the meta-analysis pipeline. The necessary scripts are located in the metaanalysis folder provided by metaGWASmanager, and they should be run here:
"/storage/consortium_name/scripts/metaanalysis"
J) 01-locate-input-files.sh
First, CAs must prepare the input-file-list.txt containing all the study names to be meta-analyzed as follows:
| GCKD_2022-10-04 |
| Study_name_2 |
| Study_name_3 |
| ... |
| Study_name_N |
The file should be saved within the created metaanalysis directory.
"/storage/consortium_name/metaanalysis"
Given a phenotype (e.g., uacr_int), the 01-locate-input-files.sh script creates the input-files-with-path.txt file. It contains the full paths to the cleaned association-output files across all studies intended for meta-analysis.
bash 01-locate-input-files.sh uacr_int
K) 02-prepare-ma-input.HQ.sh and 02-prepare-ma-input.LQ.sh
They are both mostly the same scripts, but the first with stringent filter settings and second with lenient filter settings. Based on the filter configurations, input files per specific phenotype and study will be generated in the input folder.
"/storage/consortium_name/metaanalysis/uacr_int/HQ/input"
Command line to run:
bash 02-prepare-ma-input.HQ.sh uacr_int
L) 03-prepare-metal-params.sh
The 03-prepare-metal-params.sh script uses the metal-params.txt template, which contains different filters (MAF, MAC, genomic control [yes, or not], etc.) to be set by CAs according to their preferences. The selected options are then used by the METAL software for conducting pior meta-analysis steps (e.g., genomic control correction) and meta-analysis.
To run the script, the phenotype (e.g., uacr_int) and type of selected variants (e.g, LQ or HQ) must be provided as arguments:
bash 03-prepare-metal-params.sh uacr_int HQ
By running this script, the configured metal-params.txt, needed to perform meta-analysis, will be generated at:
"/storage/consortium_name/metaanalysis/uacr_int/HQ/metal_output_HQ"
M) 04-run-metal.sh
Finally, run meta-analysis as follow:
bash 04-run-metal.sh uacr_int HQ
Results will be stored in a output folder:
"/storage/consortium_name/metaanalysis/uacr_int/HQ/metal_output_HQ/output"
Here CAs will find a .info file, encompassing details about the meta-analysis process, parameters and input files used. Additionally, a .tbl file containing the meta-analysis results will be generated.
| MarkerName | Allele1 | Allele2 | Freq1 | FreqSE | MinFreq | MaxFreq | Effect | StdErr | P-value | Direction | HetISq | HetChiSq | HetDf | HetPVal | n_total_sum | n_effective_sum | mac_sum | oevar_imp_sum |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 3:53066069:C:G | c | g | 0.9993 | 0.0000 | 0.9993 | 0.9993 | -2.59795000 | 3.27857070 | 0.4281 | ?- | 0.0 | 0.000 | 0 | 1 | 1417 | 1417 | 2 | 1 |
| 3:100796605:A:G | a | g | 0.9936 | 0.0003 | 0.9929 | 0.9937 | 0.10832363 | 0.10150211 | 0.2859 | ++ | 0.0 | 0.502 | 1 | 0.4787 | 6367 | 6363.7 | 81.9882 | 1.99878 |
| 1:174816432:C:T | t | c | 0.0051 | 0.0015 | 0.0041 | 0.0073 | -0.08297466 | 0.12499083 | 0.5068 | -- | 0.0 | 0.557 | 1 | 0.4554 | 6367 | 5700.18 | 60.9216 | 1.74315 |
| 7:108043353:A:C | a | c | 0.0002 | 0.0000 | 0.0002 | 0.0002 | 0.37409000 | 0.63925300 | 0.5584 | +? | 0.0 | 0.000 | 0 | 1 | 4950 | 4950 | 2 | 1 |
| 2:234097095:A:G | a | g | 0.1593 | 0.0128 | 0.1519 | 0.1814 | -0.00944623 | 0.02174086 | 0.6639 | -+ | 0.0 | 0.166 | 1 | 0.6836 | 6367 | 6363.68 | 2018.28 | 1.99803 |
| 5:128938723:G:T | t | g | 0.2990 | 0.0011 | 0.2968 | 0.2995 | -0.00445147 | 0.01767802 | 0.8012 | +- | 68.3 | 3.157 | 1 | 0.0756 | 6367 | 6343.99 | 3806.78 | 1.98925 |
| 8:2498343:C:T | t | c | 0.0002 | 0.0000 | 0.0002 | 0.0002 | 0.17083500 | 0.64209000 | 0.7902 | +? | 0.0 | 0.000 | 0 | 1 | 4950 | 4562.53 | 2.15686 | 0.921723 |
More information about the meta-analysis process here.