Last updated: 2020-02-04

Checks: 5 2

Knit directory: PSYMETAB/

This reproducible R Markdown analysis was created with workflowr (version 1.6.0). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.

R Markdown file: uncommitted changes

The R Markdown file has unstaged changes. To know which version of the R Markdown file created these results, you’ll want to first commit it to the Git repo. If you’re still working on the analysis, you can ignore this warning. When you’re finished, you can run wflow_publish to commit the R Markdown file and build the HTML.

Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

Seed: set.seed(20191126)

The command set.seed(20191126) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

Session information: recorded

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

Cache: none

Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.

File paths: absolute

Using absolute paths to the files within your workflowr project makes it difficult for you and others to run your code on a different machine. Change the absolute path(s) below to the suggested relative path(s) to make your code more reproducible.

absolute	relative
/data/sgg2/jenny/projects/PSYMETAB	.

Repository version: 8591fcc

Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility. The version displayed above was the version of the Git repository at the time these results were generated.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated:


Ignored files:
    Ignored:    .drake/
    Ignored:    analysis/GRS/
    Ignored:    analysis/GWAS/
    Ignored:    analysis/QC/
    Ignored:    data/processed/
    Ignored:    data/raw/
    Ignored:    packrat/lib-R/
    Ignored:    packrat/lib-ext/
    Ignored:    packrat/lib/

Untracked files:
    Untracked:  ._docs
    Untracked:  .future/
    Untracked:  GWAS.out
    Untracked:  PSYMETAB_GWAS.log
    Untracked:  Rplots.pdf
    Untracked:  analysis/._data_processing_in_genomestudio.Rmd
    Untracked:  analysis/._quality_control.Rmd
    Untracked:  cache_log.csv
    Untracked:  debug/
    Untracked:  init_analysis.future.out
    Untracked:  post_impute_qc.out
    Untracked:  qc_prep.log

Unstaged changes:
    Modified:   analysis/quality_control.Rmd
    Deleted:    packrat/snapshot.lock
    Modified:   pre_imputation_qc.log
    Modified:   pre_impute_qc.out

Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.

These are the previous versions of the R Markdown and HTML files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view them.

File	Version	Author	Date	Message
Rmd	6dcbccf	Sjaarda Jennifer Lynn	2020-02-03	update rmd files
Rmd	e6f7fb5	Jenny	2019-12-17	improve website
Rmd	82159d6	Sjaarda Jennifer Lynn	2019-12-10	run meta, prepare for grs
Rmd	dd4a16b	Jenny	2019-12-09	misc project updates re main gwas analysis
html	46477dd	Jenny Sjaarda	2019-12-06	Build site.
Rmd	b503ef0	Sjaarda Jennifer Lynn	2019-12-06	add more details to website
html	b6cb027	Jenny Sjaarda	2019-12-06	Build site.
Rmd	487b5f5	Sjaarda Jennifer Lynn	2019-12-06	update website, add qc description

The following document outlines and summarizes the quality control and processing procedure that was followed to create a clean, imputed dataset.

Step 1: Prepare and cluster genomestudio files.

Step 1 was performed entirely on CHUV computer

Part A: Randomize IDs.

Genetic sampleIDs were recoded according to GPCR algorithm to ensure genetic participants are not identifiable.
code/radomize_IDs.r was run on CHUV computer before building GenomeStudio project.
Creates a new csv file which was used to create a GenomeStudio project with data provided by lab in Geneva.
Requires manual addition of header before uploading to GenomeStudio.

[Header],,,,,,,,,,,,,
Investigator Name,,,,,,,,,,,,,
Project Name,,,,,,,,,,,,,
Experiment Name,,,,,,,,,,,,,
Date,,,,,,,,,,,,,
[Manifests],,,,,,,,,,,,,
A,GSA_UPPC_20023490X357589_A1,,,,,,,,,,,,
[Data],,,,,,,,,,,,,

Some samples were found to be duplicates (i.e. 2 samples at 2 different time points were analyzed for the same individual) and they were recoded to have ID as: ${ID}002.

Part B: Create GenomeStudio files.

Instructions can be found here.
Required files:
- Sample sheet: as csv file (created above).
- Data repository: as idat files.
- Manifest file: as bpm file.
- Cluster file: as egt file.
Data provided from Mylene Docquier, copied from sftp and saved here: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS\data.
Create new IDs based on GPCR randomization (see code/randomize_IDs.r), and save to above folder as: Eap0819_1t26_27to29corrected_7b9b_randomizedID.csv.
Note that original IDs can be found in the same folder at the file: Eap0819_1t26_27to29corrected_7b9.csv, if needed.
Create empty folder here: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS, named: GS_project_26092019 (data of creation).
Using new IDs, create genome studio project as follows:
1. Open GenomeStudio.
2. Select: File > New Genotyping Project.
3. Select L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS as project repository.
4. Under project name: use “GS_project_26092019” and click “Next”.
5. Select “Use sample sheet to load intensities” and click “Next”.
6. Select sample, data and manifests as specified below and click “Next”:
  - Sample sheet: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS\data\Eap0819_1t26_27to29corrected_7b9b_randomizedID.csv,
  - Data repository: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS\data,
  - Manifest repository: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS\data.
7. Select “Import cluster positions from cluster file” and choose cluster file located here: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS\data\GSPMA24v1_0-A_4349HNR_Samples.egt and click “Finish”.
Genome studio files were created using the following required files:
- Sample sheet: as csv file (created above).
- Data repository: as idat files (provided from Mylene Docquier).
- Manifest file: as bpm file (provided from Mylene Docquier).
- Cluster file: as egt file (provided from Mylene Docquier).
For detailed description see Data Processing in Genome Studio (section: Creating GenomeStudio file)

Part C: Clustering and PLINK conversion.

Data saved on CHUV servers at the following location: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS\GS_project_26092019.
GS_project_26092019.bsc was opened (requires Genome Studio) and used for clustering.
Clustering was performed according to the guidelines in the webinar (see notes on creating custom cluster files) with the following procedure:
1. Cluster: cluster all SNPs, evaluate samples.
  1. Once data was opened, all SNPs were clustered by clicking “Cluster all SNPs button” in the top panel (icon with 3 red, purple, blue ovals). SNP statistics and heritability estimate updates were ignored.
  2. In the samples table (bottom left panel) call rate was recalculated by clicking “Calculate” (calculator icon) to recalculate SNP call rates with new cluster positions. SNP statistics and heritability estimate updates were ignored.
  3. The sample table was then sorted by the “Call Rate” column and samples with call rate <95% were selected and removed from downstream processing (right click, select “Exclude Selected Samples”), 8 samples fell below this cut-off. Updates were ignored. In “SNP Graph” right click and deselect “Show Excluded Samples”.
2. Recluster: cluster sex chromosomes, cluster autosomes.
  1. “SNP Table” was filtered by (“Chr” = Y) using filter icon.
  2. Using the 3rd and 7th SNP (index 3010 and 3014, 3 samples had missing values for 3010 and sex was determined to be female using 3014), males were selected as those with high intensities (females have no Y chromosome) and set to have an aux value of 101 by right clicking the selected samples in the “Samples Table”. Similarly, females were selected and set to have an aux value of 102.
  3. “Samples Table” was filtered to have (“Aux” > 100).
  4. “Samples Table” was sorted on “Aux” column and females were selected (value 102) and excluded. Updates were ignored.
  5. In “SNP Table”, all SNPs were selected and Y-snps were clustered (right click and “Cluster Selected SNPs”) on only male samples. Updates were ignored.
  6. Females were re-added to the project in the “Samples Table”.
  7. “SNP Table” was filtered by (“Chr” = X) using filter icon.
  8. In “Samples Table”, males were selected (Aux value 101) and excluded. Updates were ignored.
  9. In “SNP Table”, all SNPs were selected and X-SNPs were clustered (right click and “Cluster Selected SNPs”) on only female samples. Updates were ignored.
  10. Males were re-added to the project in the “Samples Table”.
  11. “SNP Table” was filtered by [ !(“Chr” = X ) AND !(“Chr” = Y ) ], using filter icon.
  12. In “SNP Table”, all SNPs were selected and autosomal SNPs were clustered (right click and “Cluster Selected SNPs”) on all good quality samples. Update SNP statistics.
  13. In the samples table (bottom left panel) call rate was recalculated by clicking “Calculate” (calculator icon) to recalculate SNP call rates with new cluster positions. SNP statistics and heritability estimate updates were ignored.
  14. SNP statistics were updated (“Analysis” > “Update SNP statistics”).
3. Review and edit: use filters and scores to evaluate SNPs, correct or zero SNPs as needed. No manual editing was performed.
New genome studio was exported to the following location: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS, and named: GS_project_26092019_cluster.
Remove filtered individuals:
- In samples table, remove filter for “Aux” > 100.
- Recalculate SNP statistics for these 8 samples only (to save time).
- Note that sample call rates have increased with new clustering positions.
Project was exported as PLINK file to the following location: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS, and named: PLINK_091019_0920.
Individual PLINK files within above folder were named according to parent directory as: PSYMETAB_GWAS.ped and PSYMETAB_GWAS.map.
Ultimately, the final ped/map files reside here: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS\PLINK_091019_0920.

Notes on GenomeStudio.

Processing data within GS software is slow in comparison to the rest of the pipeline.
There is no command line version, only a licensed, GUI version availabe.
It is impossible to fully automate and/or parallize.

Data updates and releases.

Initial data was received on April 8, 2019:
- Final two plates were received on May 17, 2019.
- Processing began with initial files.
July 18, 2019 update:
- It came to our attention that 15 participants were genotyped that did not consent.
- This list was sent to Mylene to be removed.
- Plates 27 to 29 were re-provided on 08/08/2019 without these 15 individuals (list below, and provided by Severine in email).

013CB
017CB
074CB
095CB
150CRV
192CRV
193CRV
156CSM
181CSM
191CSM
224UAS
234GL
058GP
246GP
089PP

Until new file was provided, these participants were removed using PLINK to avoid any further analysis of these individuals.
The new genomestudio file was copied to: L:\PCN\UBPC\ANALYSES_RECHERCHE\Jenny\PSYMETAB_GWAS\PSYMETAB_GS2\Plates27to29_0819.
The same process above was followed (data opened in GS, cluster positions imported, and data saved to Plates27to29_0819_cluster, and PLINK_270819_0457).
Old files were deleted to remove all data containing these individuals.
Updated files were then copied to PSYMETAB_GS1.
August 28, 2019 update:
- Mylene provided one single genome studio file with all samples (excluding the list from Severine).
- These files were copied to UPPC folders and custom clustering was re-performed.
- These changes are reflected in the above description.
- All old files were subsequently deleted to ensure the data from these participants is completely removed from all databases.
- As of September 3, 2019, all clustering was complete and final PLINK files (PLINK_030919_0149) were copied to SGG directory (names of plink files according to parent directory: DATA).
September 6, 2019 update:
- It was decided that all IDs part of PSYMETAB should be randomized to ensure they are not identifiable.
- We had a meeting to discuss (Celine, Fred, Chin, Nermine, and Claire), and decided to use a CHUV program (GPCR) for the randomization process.
- We requested with Mylene to create a new project with the new IDs, but she suggested to create our own GS project.
- She provided all relevant data to create our own GS project.
- The description above reflects these changes.
As of October 11, all GS file were created, clustered and exported as PLINK files and subsequently moved to the sgg server.

Part D: Copy files to SGG server.

PLINK files were copied to SGG servers using FileZilla
Host name: je4649@hpc1.chuv.ch
Password: <chuv-password>
Port: 22
Output saved to: /data/sgg2/jenny/projects/PSYMETAB_GWAS/data/raw.

All subsequent steps were performed on the sgg server and run using drake plan

Step 2: Pre-quality control data prep.

Processed sex and ethnicity files to be used in QC scripts.
Sex file was created according to the input specified on plink man page (FID, IID, sex [M/F]).
Ethnicity input file to be used in R script for comparison to genetically derived ethnic groups (by snpWeights).
Recodes ethnic groups as follows:
- Changes French codes to English.
- Changes missing to unknown.
- Groups small ethnic groups to missing.
A1 rsid conversion file was updated to remove all SNPs labeled with a [.] (see data sources).
Create duplicates file

Step 3: Pre-imputation quality control.

Results of Step 3-5 are saved to analysis/QC Each step corresponds to one folder within analysis/QC

0. Preprocessing.

Create binary PLINK files (if necessary):
- 761641 variants initially.
- 2767 individuals initially.
Exclude Y and MT variants:
- 11641 Y / MT variants removed.
- 750000 variants remaining.
Remove duplicate variants:
- 7945 duplicates removed.
- 742055 variants remaining.
Update sex:
- Using the file located at: data/processed/phenotype_data/PSYMETAB_GWAS_sex.txt (created above).
```
   F    M 
1298 1469 
```
Remove duplicate individuals that were identified previously (i.e. two different IDs for the same participant):
- 2 duplicates identified and removed
- 2752 indviduals remaining.
Sanity check: ensure that duplicates have the same genetic info"
- From the PLINK webpage: Note that KING kinship coefficients are scaled such that duplicate samples have kinship 0.5, not 1. First-degree relations (parent-child, full siblings) correspond to ~0.25, second-degree relations correspond to ~0.125, etc.
- If all duplicate ID samples have KINSHIP ~0.5, then indeed they are genetic duplicates.

   #FID1         ID1 FID2      ID2   NSNP   HETHET          IBS0  KINSHIP
1   2071 BEEEDIGO002  224 BEEEDIGO 703110 0.178137 0.00000000000 0.499539
2   1873 CQLIXEZP002   64 CQLIXEZP 703045 0.153413 0.00000142238 0.499504
3   1965 EFWKQOIK002 1433 EFWKQOIK 697403 0.151525 0.00000860335 0.496680
4   1886 HFNWJHCI002 1448 HFNWJHCI 702845 0.153089 0.00000426837 0.499547
5   2075 HROOJNCI002  553 HROOJNCI 702167 0.155970 0.00000284833 0.499257
6   1974 IOAWLZGK002  549 IOAWLZGK 704278 0.153028 0.00000000000 0.499847
7   2314 KLFEBCIE002 1916 KLFEBCIE 700799 0.153949 0.00000570777 0.499007
8   2073 LWCGLSDP002  317 LWCGLSDP 702226 0.150114 0.00000427213 0.499363
9   2379 PBAIFEMQ002 2070 PBAIFEMQ 700642 0.154083 0.00000285452 0.498820
10  2009 PNWDYVRH002  494 PNWDYVRH 703993 0.153736 0.00000284094 0.499806
11  2068 QHNUPGWK002  318 QHNUPGWK 702891 0.154500 0.00000569078 0.499363
12  1928 QZAUHIPY002  559 QZAUHIPY 702896 0.144711 0.00000142269 0.499826
13  2067 SSITXXAY002  283 SSITXXAY 702409 0.152603 0.00000284734 0.499418
14  1947 WKBFDWJF002  566 WKBFDWJF 703642 0.153506 0.00000284235 0.499783
15  1657 XABRILAR002 1385 XABRILAR 698282 0.154672 0.00000000000 0.497315

1. Strand alignment (so all SNPs are on positive strand).

Update chromosome
Update position
Flip alleles on negative strand
Extract SNPs that are not in strand file
Remove any non autosomal / X chromosomal variants (might have changed due to the pos and chr update):
- 11 duplicates removed.
- 741937 variants remaining.
Update rsids using data/processed/reference_files/rsid_conversion.txt
Remove duplicate variants:
- 6723 duplicates removed.
- 734328 variants remaining.

2. Removal of SNPs that have MAF zero.

Calculate frequency of all SNPs
Remove MAF 0 SNPs:
- 93578 variants with MAF = 0.
- 640750 variants remaining.

3. Missingness.

Exclude variants with >10% missingness (using geno --0.1):
- 6480 variants removed.
- 634270 variants remaining.
Exclude individuals with >10% missingness (using mind --0.1):
- 3 individuals removed.
  - 2749 individuals remaining.
Exclude variants with >5% missingness (using geno --0.05):
- 8256 variants removed.
- 626014 variants remaining.
Exclude individuals with >5% missingness (using mind --0.05):
- 3 individuals removed.
- 2746 individuals remaining.
Exclude variants with >1% missingness (using geno --0.01):
- 35957 variants removed.
- 590057 variants remaining.
Exclude individuals with >1% missingness (using mind --0.01):
- 5 individuals removed.
- 2741 individuals remaining.

Total removed: 50693 variants (7.91%) and 11 individuals (0.40%).

4. Sex check.

Perform sex check.
Remove unambiguous sex violations.
- 26 individuals removed.
- 2715 individuals remaining.

5. Imputation preparation.

Write frequency of final QC’d file (from step 4) to file (using -- freq).
Using McCarthy Group Tools, QC’d files were prepared for imputation using the script HRC-1000G-check-bim-NoReadKey.pl (download link).
- This script checks: Strand, alleles, position, Ref/Alt assignments and frequency differences.
- Produces: A set of plink commands to update or remove SNPs based on the checks as well as a file (FreqPlot) of cohort allele frequency vs reference panel allele frequency.
- Updates: Strand, position, ref/alt assignment.
- Removes: A/T & G/C SNPs if MAF > 0.4, SNPs with differing alleles, SNPs with > 0.2 allele frequency difference (can be removed/changed in V4.2.2), SNPs not in reference panel
Replace underscores in fam file since vcf conversion uses understcores between FID and IID.
Run the generated Run-plink.sh script from #2.
Sort outputed vcf files.
Download zipped files to personl or CHUV files to copy to Michigan Imputation Server.

Step 4: Imputation:

6. Run and download imputation.

Upload downloaded QC’d vcf.gz files to Michigan Imputation Server as follows:
- Select Run, Genotype Imputation (Minimac4).
- Reference panel: HRC r1.1 2016 (GRCh37/hg19).
- Array build: GRCh37/hg19.
- rsq Filter: off.
- Phasing: Eagle v2.4 (phased output).
- Population: EUR.
- Mode: Quality Control & Imputation.

Download imputation, using password from email retrieve the following files:
- QC report.
- QC stats.
- Logs.
- Imputation results.

   Chr Num imputed variants
1    1              3069931
2    2              3392237
3    3              2821894
4    4              2787581
5    5              2588168
6    6              2460111
7    7              2289305
8    8              2242705
9    9              1686471
10  10              1927503
11  11              1936990
12  12              1848117
13  13              1385433
14  14              1270436
15  15              1139215
16  16              1281297
17  17              1090072
18  18              1104755
19  19               868554
20  20               884983
21  21               531276
22  22               524544
23 all             39131578

7. Check imputation.

Using Will Rayner’s post imputation tool, produce a visual summary of the imputated results from the Michigan Imputation Server:
- Imputation report.
- Cross cohort report.

Step 5: Post imputation quality control

8. PLINK conversion

Remove SNPs with info < 0.30

   Chr Num imputed variants R2 filtered
1    1              3069931     2223452
2    2              3392237     2454557
3    3              2821894     2059285
4    4              2787581     2043343
5    5              2588168     1887050
6    6              2460111     1813223
7    7              2289305     1659374
8    8              2242705     1632513
9    9              1686471     1218207
10  10              1927503     1411585
11  11              1936990     1421749
12  12              1848117     1352003
13  13              1385433     1019900
14  14              1270436      920528
15  15              1139215      816986
16  16              1281297      894859
17  17              1090072      771058
18  18              1104755      798012
19  19               868554      607422
20  20               884983      631946
21  21               531276      374376
22  22               524544      367153
23 all             39131578    28378581

Update bim file to include rsIDs instead of chr:bp convention:
- Output of Michigan Impuation server is in format chr:bp:ref:alt.
- Convenient to have SNPs in regular rsIDs for extraction, etc.
- If there is no known rsID, SNP name is left as chr:bp:ref:alt.
- Reference file for determining rsIDs can be found here: /data/sgg2/jenny/data/dbSNP/dbSNP_SNP_list_chr${chr}.txt, which was processed according to description in jenny/SGG_generic/scripts/public_data.sh.

9. Extract typed SNPs

Write a list of only genotypd data to run in snpweights software (using require=info "TYPED" flag).

   Chr Num imputed variants R2 filtered Typed SNPs
1    1              3069931     2223452      44252
2    2              3392237     2454557      45514
3    3              2821894     2059285      37277
4    4              2787581     2043343      34139
5    5              2588168     1887050      32223
6    6              2460111     1813223      38957
7    7              2289305     1659374      30649
8    8              2242705     1632513      28611
9    9              1686471     1218207      23815
10  10              1927503     1411585      27714
11  11              1936990     1421749      27722
12  12              1848117     1352003      26603
13  13              1385433     1019900      19346
14  14              1270436      920528      17920
15  15              1139215      816986      17002
16  16              1281297      894859      18447
17  17              1090072      771058      16646
18  18              1104755      798012      15625
19  19               868554      607422      12997
20  20               884983      631946      13557
21  21               531276      374376       7582
22  22               524544      367153       8104
23 all             39131578    28378581     544702

Hard call back to genotypes (bim, bed, and fam files) using the --hard-call-threshold flag set at 0.1.
Merge by chromosome genotyped data into single file.

10. Merge imputed SNPs

Convert fileset from step #8 back to vcf as there is no merge function in plink (using the flag --recode vcf id-paste=iid vcf-dosage=HDS).
Merge using bcftools concat.
Convert back to pgen using plink2 --vcf <output-name> dosage=HDS.

11. Relatedness

Calculate missingness.
Remove subjects and individuals with missingness >10%.
- 1184212 variants removed.
- 0 individuals removed.
Compute KING table (--make-king-table).

Identify duplicate individuals using kinship threshold > 0.35 as recommended by PLINK.

#FID1	ID1	FID2	ID2	NSNP	HETHET	IBS0	KINSHIP
1353_WIELRZDD	1353_WIELRZDD	1140_FJNJEXCM	1140_FJNJEXCM	26566711	0.0685663	0.0000000	0.499473
1457_ABDJGAXW	1457_ABDJGAXW	104_ZXDVAJUD	104_ZXDVAJUD	27026534	0.0515445	0.0000000	0.499956
1821_0030GE	1821_0030GE	628_SJJCZOXT	628_SJJCZOXT	27036506	0.0520661	0.0000000	0.499937
1937_OINDBCQM	1937_OINDBCQM	1936_SONUEGRB	1936_SONUEGRB	27021947	0.0516029	0.0000001	0.499964
2020_JNQSZGGL	2020_JNQSZGGL	1305_XTVNZTRY	1305_XTVNZTRY	26975108	0.0526075	0.0000000	0.499941
2084_JO001	2084_JO001	220_UUHDCMNL	220_UUHDCMNL	27017927	0.0512845	0.0000000	0.499938
2148_JO280	2148_JO280	2146_JO276	2146_JO276	27011980	0.0519508	0.0000001	0.499942
2187_S008	2187_S008	1298_YLXVRYDT	1298_YLXVRYDT	27014700	0.0535244	0.0000000	0.499975
2239_S060	2239_S060	1851_0062GE	1851_0062GE	26706894	0.0543626	0.0000003	0.499584
2271_S092	2271_S092	1813_0021GE	1813_0021GE	26980864	0.0534746	0.0000000	0.499881
2272_S093	2272_S093	2174_JO426	2174_JO426	27029496	0.0521779	0.0000000	0.499968
2637_139CSM	2637_139CSM	2302_HLHILECG	2302_HLHILECG	27037160	0.0520094	0.0000000	0.499967
2678_184CSM	2678_184CSM	2042_WXPQPUWI	2042_WXPQPUWI	27013111	0.0525789	0.0000000	0.499924
2689_198CSM	2689_198CSM	509_CHMLAUKH	509_CHMLAUKH	26954876	0.0518025	0.0000002	0.499894
2693_202CSM	2693_202CSM	1955_OQLLRBMM	1955_OQLLRBMM	27019285	0.0515285	0.0000000	0.499944
2760_274UAS	2760_274UAS	295_YGGZRBYJ	295_YGGZRBYJ	27047919	0.0523141	0.0000000	0.499972

A set of 16 duplicate particpants were found that were not expected.
These duplicates needed to be manually checked in database with Celine (see below).

Manually determine cause for duplicates.
- These pairs of individuals were found to be in both studies under different IDs (PSYMETAB and Severines):
```
# 139CSM  HLHILECG
# 184CSM  WXPQPUWI
# 198CSM  CHMLAUKH
# 202CSM  OQLLRBMM
# 274UAS  YGGZRBYJ
# JO001 UUHDCMNL
# S008  YLXVRYDT
# S092  0021GE
```
- These pairs of individuals are duplicates but with different codes and we are keeping the one on the right:
```
# ABDJGAXW  ZXDVAJUD
# JO280 JO276
# S093  JO426
```
- These pair of individuals are not the same and the genetic info matches the one on the right (remove left):
```
# JNQSZGGL  XTVNZTRY
# WIELRZDD  FJNJEXCM
# OINDBCQM  SONUEGRB
```
- These pair of individuals could not be properly sorted and are therefore both being deleted:
```
# S060  0062GE
# 0030GE    SJJCZOXT
```
Create file of duplicates to be removed (based on above criteria).
Remove duplicates.
- 2715 individuals originally.
- 18 individuals removed.
- 2697 individuals remaining.
Recalculate kinship and identify remaining related individauls using kinship threshold > 0.0884) as recommended by KING authors.
- #r countLines(paste0("analysis/QC/10_merge_imputed/", study_name, "2.related.filter.smiss"))[1] - 1 related pairs identified.
- #r countLines(paste0("analysis/QC/10_merge_imputed/", study_name, "2.duplicates"))[1] - 1 duplicate pairs identified.
Determine maximal set of unrelated individuals based on kinship results.
- This was done using personal script (code/qc/relatedness_filter.r).
- PLINK does not guarantee maximum number of samples (see here: Relationship-based pruning), where it states: PLINK tries to maximize the final sample size, but this maximum independent set problem is NP-hard, so we use a greedy algorithm which does not guarantee an optimal result.
Remove related individuals (note that if using mixed model design we can include all the related IDs, but remove the problematic duplicates).
- 50 related IDs removed.
- 2647 individuals remaining.

12. Ethnicty check and admixture estimation

13. HWE check

14. MAF check

15. Final processing

Code

code was run in 5 steps

sessionInfo()

R version 3.5.3 (2019-03-11)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: CentOS Linux 7 (Core)

Matrix products: default
BLAS: /data/sgg2/jenny/bin/R-3.5.3/lib64/R/lib/libRblas.so
LAPACK: /data/sgg2/jenny/bin/R-3.5.3/lib64/R/lib/libRlapack.so

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] tidyselect_0.2.5        kableExtra_1.1.0        R.utils_2.9.2          
 [4] R.oo_1.23.0             R.methodsS3_1.7.1       TwoSampleMR_0.4.25     
 [7] reader_1.0.6            NCmisc_1.1.6            optparse_1.6.4         
[10] readxl_1.3.1            ggthemes_4.2.0          tryCatchLog_1.1.6      
[13] futile.logger_1.4.3     DataExplorer_0.8.0      taRifx_1.0.6.1         
[16] qqman_0.1.4             MASS_7.3-51.5           bit64_0.9-7            
[19] bit_1.1-14              rslurm_0.5.0            rmeta_3.0              
[22] devtools_2.2.1          usethis_1.5.1           data.table_1.12.8      
[25] clustermq_0.8.8.1       future.batchtools_0.8.1 future_1.15.1          
[28] rlang_0.4.2             knitr_1.26              drake_7.9.0.9000       
[31] forcats_0.4.0           stringr_1.4.0           dplyr_0.8.3            
[34] purrr_0.3.3             readr_1.3.1             tidyr_1.0.0            
[37] tibble_2.1.3            ggplot2_3.2.1           tidyverse_1.3.0        
[40] pacman_0.5.1            processx_3.4.1          workflowr_1.6.0        

loaded via a namespace (and not attached):
 [1] backports_1.1.5      plyr_1.8.5           igraph_1.2.4.2      
 [4] lazyeval_0.2.2       storr_1.2.1          listenv_0.8.0       
 [7] digest_0.6.23        htmltools_0.4.0      fansi_0.4.1         
[10] magrittr_1.5         checkmate_1.9.4      memoise_1.1.0       
[13] base64url_1.4        remotes_2.1.0        globals_0.12.5      
[16] modelr_0.1.5         prettyunits_1.1.0    colorspace_1.4-1    
[19] rvest_0.3.5          rappdirs_0.3.1       haven_2.2.0         
[22] xfun_0.11            callr_3.4.0          crayon_1.3.4        
[25] jsonlite_1.6         zeallot_0.1.0        brew_1.0-6          
[28] glue_1.3.1           gtable_0.3.0         webshot_0.5.2       
[31] pkgbuild_1.0.6       scales_1.1.0         futile.options_1.0.1
[34] DBI_1.1.0            Rcpp_1.0.3           viridisLite_0.3.0   
[37] progress_1.2.2       txtq_0.2.0           htmlwidgets_1.5.1   
[40] httr_1.4.1           getopt_1.20.3        calibrate_1.7.5     
[43] ellipsis_0.3.0       pkgconfig_2.0.3      dbplyr_1.4.2        
[46] reshape2_1.4.3       later_1.0.0          munsell_0.5.0       
[49] cellranger_1.1.0     tools_3.5.3          cli_2.0.1           
[52] generics_0.0.2       broom_0.5.3          evaluate_0.14       
[55] yaml_2.2.0           fs_1.3.1             packrat_0.5.0       
[58] nlme_3.1-143         whisker_0.4          formatR_1.7         
[61] proftools_0.99-2     xml2_1.2.2           compiler_3.5.3      
[64] rstudioapi_0.10      filelock_1.0.2       testthat_2.3.1      
[67] reprex_0.3.0         stringi_1.4.5        highr_0.8           
[70] ps_1.3.0             desc_1.2.0           lattice_0.20-38     
[73] vctrs_0.2.1          pillar_1.4.3         lifecycle_0.1.0     
[76] networkD3_0.4        httpuv_1.5.2         R6_2.4.1            
[79] promises_1.1.0       gridExtra_2.3        sessioninfo_1.1.1   
[82] codetools_0.2-16     lambda.r_1.2.4       assertthat_0.2.1    
[85] pkgload_1.0.2        rprojroot_1.3-2      withr_2.1.2         
[88] batchtools_0.9.12    parallel_3.5.3       hms_0.5.3           
[91] grid_3.5.3           rmarkdown_1.18       git2r_0.26.1        
[94] lubridate_1.7.4

Data quality control and processing