These six options can be added to most commands:
--extract <extractfile> - tells LDAK to use only the predictors specified in <extractfile>. Note that <extractfile> is expected to have one column, which should provides predictor names (if the file contains more than one column, only the first will be used).
--exclude <excludefile> - tells LDAK to not use the predictors specified in <excludefile>. Note that <excludefile> is expected to have one column, which should provides predictor names (if the file contains more than one column, only the first will be used). Note also that the filtering specified using --exclude takes priority over the filtering specified using --extract.
--chr <integer> - tells LDAK to only use predictors on a particular chromosome. If you use --chr AUTOSOMES, LDAK will only use Chromosomes 1-22 (in humans, these are the autosomes), while if you use --chr ODD or --chr EVEN, LDAK will only use odd or even chromosomes.
--snp <predname> - tells LDAK to only use the named predictor.
--keep <keepfile> - tells LDAK to only use the samples specified in <keepfile>. Note that <keepfile> is expected to have two columns which should provides the FIDs and IIDs of samples (if the file contains more than two columns, only the first two will be used).
--remove <removefile> - tells LDAK to not use the samples specified in <removefile>. Note that <removefile> is expected to have two columns which should provides the FIDs and IIDs of samples (if the file contains more than two columns, only the first two will be used). Note also that the filtering specified using --remove takes priority over the filtering specified using --keep.
In addition, for many commands, you can add --pheno <phenofile> and LDAK will only consider samples for which phenotypes are available.
Lastly, a few commands (e.g., LDAK-KVIK, Single-Predictor Analysis, Elastic Predict, Calculate Kinships and Make Data) allow for on-the-fly predictor quality control using --min-maf <float>, --max-maf <float>, --min-obs <float>, --min-var <float> and/or --min-info <float>. For other commands, you should first use Calculate Statistics to identify which predictors pass quality control, then restrict to these using --extract
If you would like to filter predictors or individuals based on quality control metrics (e.g., minor allele frequency, variance, missingness or information score), it is generally necessary to first generate lists of predictors and/or individuals that pass the chosen filters (e.g., using the results from Calculate Statistics), then provide these lists using --extract and/or --keep. However, for a few commands (e.g., LDAK-KVIK, Single-Predictor Analysis, Elastic Predict and Calculating Kinships), it is possible to filter predictors on-the-fly using --min-maf <float>, --max-maf <float>, --min-obs <float>, --min-var <float> and/or --min-info <float>.
