filter_GWAS {QCGWAS}

Automated filtering and reformatting of GWAS results files

Description

This function was created as a convenient way to automate the removal of low-quality and non-autosomal SNPs. It includes the same formatting options as QC_GWAS.

Usage

filter_GWAS(ini_file,
            GWAS_files, output_names,
            gzip_output = TRUE,
            dir_GWAS = getwd(), dir_output = dir_GWAS,
            FRQ_HQ = NULL, HWE_HQ = NULL,
            cal_HQ = NULL, imp_HQ = NULL,
            FRQ_NA = TRUE, HWE_NA = TRUE,
            cal_NA = TRUE, imp_NA = TRUE,
            ignore_impstatus = FALSE,
            remove_X = FALSE, remove_Y = FALSE,
            remove_XY = FALSE, remove_M = FALSE,
            header_translations,
            check_impstatus = FALSE,
            imputed_T = c("1", "TRUE", "yes", "YES", "y", "Y"),
            imputed_F = c("0", "FALSE", "no", "NO", "n", "N"),
            imputed_NA = NULL,
            column_separators = c("\t", " ", "", ",", ";"),
            header = TRUE, nrows = -1, nrows_test = 1000,
            comment.char = "", na.strings = c("NA", "."),
            out_header = "original", out_quote = FALSE,
            out_sep = "\t", out_eol = "\n", out_na = "NA",
            out_dec = ".", out_qmethod = "escape",
            out_rownames = FALSE, out_colnames = TRUE, ...)

Arguments

ini_file

(the filename of) a table listing the files to be processed and the filters to be applied. See 'Details'.

GWAS_files

character vector: when no ini_file is provided, this identifies the files to be processed. See 'Details'.

output_names

character vector: the filenames for the output files. The default option is to use the input filenames. Note that, unlike with other QCGWAS functions, the file extensions should be included (However, the function will automatically add ".gz" when the files are compressed.

gzip_output

logical; should the output files be compressed?

dir_GWAS, dir_output

character-strings specifying the directory address of the folders for the input files and the output, respectively. Note that R uses forward slash (/) where Windows uses backslash (\).

FRQ_HQ, HWE_HQ, cal_HQ, imp_HQ

Numeric vectors. When no ini_file is provided, these arguments specify the filter threshold-values for allele frequency, HWE p-value, callrate and imputation quality, respectively. Passed to HQ_filter.

FRQ_NA, HWE_NA, cal_NA, imp_NA

Logical vectors. When no ini_file is provided, these arguments specify whether missing values (of allele frequency, HWE p-value, callrates and imputation quality, respectively) are excluded (TRUE) or ignored (FALSE). Passed to HQ_filter.

ignore_impstatus

Logical vector. When no ini_file is provided, this argument specifies whether imputation status is taken into account when applying the filters. If FALSE, HWE p-value and callrate filters are applied only to genotyped SNPs, and imputation quality filters only to imputed SNPs. If TRUE, the filters are applied to all SNPs regardless of the imputation status.

remove_X, remove_Y, remove_XY, remove_M

logical; respectively whether X-chromosome, Y-chromosome, pseudo-autosomal and mitochondrial SNPs are removed. Note: these arguments accept only a single TRUE or FALSE value. Unlike the above settings, it's not possible to specify them independently for every dataset.

header_translations

translation table for column names. See translate_header for more information. If the argument is left empty, dataset is assumed to use the standard column-names used by QC_GWAS.

check_impstatus

logical; should convert_impstatus be called to convert the imputation-status column into standard values?

imputed_T, imputed_F, imputed_NA

arguments passed to convert_impstatus.

column_separators

character string or vector; specifies the values used as column delimitator in the GWAS file(s). The argument is passed to load_test; see the description of that function for more information.

nrows_test

integer; the number of rows used for "trial-loading". Before loading the entire dataset, the function load_test is called to determine the dataset's file-format by reading the top x lines, where x is nrows_test. Setting nrows_test to a low number (e.g. 150) means quick testing, but runs the risk of missing problems in lower rows. To test the entire dataset, set it to -1.

header, nrows, comment.char, na.strings, ...

arguments passed to read.table when importing the dataset.

out_header

Translation table for the column names of the output file. This argument is the opposite of header_translations: it translates the standard column-names of QC_GWAS to user-defined ones. output_header can be one of three things:

• A user specified table similar to the one used by translate_header. However, as this translates standard names into non-standard ones, the standard names should be in the right column, and the desired ones in the left. There is also no requirement for the names in the left column to be uppercase.

• The name of a file in dir_GWAS containing such a table.

• Character string specifying a standard form. See QC_GWAS, section 'Output header' for the options.

out_quote, out_sep, out_eol, out_na, out_dec, out_qmethod, out_rownames, out_colnames

arguments passed to write.table when saving the final dataset.

Details

The easiest way to use filter_GWAS is by passing an ini file to the ini_file argument. The ini file can be generated by running QC_series with the save_filtersettings argument set to TRUE. The output will include a file 'Check_filtersettings.txt', describing the (high-quality) filter settings used for each file (taking into account whether there was enough data, i.e. whether the use_threshold was met, to apply the filters).

The ini_file argument accepts both a table or the name of a file in dir_GWAS or the current R working directory.

If no ini_file is specified, the function will use the GWAS_files, x_HQ, x_NA and ignore_impstatus arguments to construct such a table. GWAS_files can either be a character vector or a single value. If a single string, all filenames containing the string will be processed. The other arguments can also be a vector or a single value; if the latter, they will be recycled to create a vector of the correct length.

If neither ini_file nor GWAS_files are specified, the function will look for a file Check_filtersettings.txt in dir_GWAS and the current R working directory.

Note that ini_file overrules the other filter settings, i.e. one cannot adjust ini_file through the other arguments.

Value

An invisible logical vector, indicating which files were successfully filtered.

Note

R is not the optimal platform for filtering GWAS files. This function was added at the request of a user, but an UNIX script is likely to be faster.