| Name | Function |
| --1 | Specify that binary phenotypes use 0/1 instead of 1/2 encoding. |
| --ac-founders | Confirm that nonfounders should be excluded from --mac/--max-mac/"--freq counts". |
| --adjust-...-field | Set --adjust-file input field names. |
| --adjust[-file] | Report basic multiple-testing adjustments for association test p-values. |
| --af-pseudocount | Add a pseudocount for each allele when estimating allele frequencies. |
| --allele1234 | Interpret/recode A/C/G/T alleles as 1/2/3/4. |
| --alleleACGT | Interpret/recode 1/2/3/4 alleles as A/C/G/T. |
| --allow-extra-chr | Explicitly permit unrecognized chromosome codes. |
| --allow-misleading-out-arg | Allow --out argument to end in a common filename extension. |
| --allow-normalize-with-split | Disable --normalize + variant-split order-of-operations warning. |
| --alt-allele | Scrape ALT alleles from a VCF or a similar file. |
| --alt1-allele | Scrape ALT1 alleles from a VCF or a similar file. |
| --autosome | Exclude all unplaced and non-autosomal variants. |
| --autosome-num | Set number of autosomal chromosomes. |
| --autosome-par | Exclude all unplaced, X, Y, and MT variants. XY/PAR1/PAR2 variants are retained. |
| --bad-freqs | Force PLINK 2 to proceed with questionable imputed allele frequencies. |
| --bad-ld | Force PLINK 2 to estimate LD from very few founders. |
| --bcf | Load BCF2 file. |
| --bed | Specify full name of input .bed file. |
| --bed-border-{bp,kb} | Extend BED intervals (for e.g. "--extract range") by the given amount. |
| --bfile | Make <prefix>.bed + .bim + .fam the main input fileset. |
| --bgen | Specify full name of input Oxford-format .bgen file. |
| --bim | Specify full name of input .bim file. |
| --bp-space | Remove variants so each pair is no closer than the given distance. |
| --bpfile | Make <prefix>.pgen + .bim + .fam the main input fileset. |
| --check-sex | Check sex assignments against those imputed from chrX/chrY genotype calls. |
| --chr | Exclude all variants not on the specified chromosomes. |
| --chr-override | Control which chromosome-set takes precedence when command line conflicts with .pvar. |
| --chr-set | Specify a nonhuman chromosome set. |
| --ci | Report confidence intervals for odds ratios. |
| --clump | Organize association reports into LD-based clumps. |
| --clump-a1-field | Specify a different --clump effect-allele field name search order. |
| --clump-allow-overlap | Permit --clump non-index variants to appear in multiple clumps. |
| --clump-bins | Adjust --clump p-value histogram bins. |
| --clump-force-a1 | Force A1 alleles to be included in --clump report for biallelic variants. |
| --clump-id-field | Specify a different --clump variant ID field name search order. |
| --clump-kb | Adjust --clump single-clump kb radius. |
| --clump-log10 | Specify -log10(p) rather than raw p-value input and/or output for --clump. |
| --clump-p-field | Specify a different --clump p-value field name search order. |
| --clump-r2 | Adjust --clump r^2 threshold. |
| --clump-range | With --clump and a region list file, report overlaps between clumps and regions. |
| --clump-range-border | Extend all --clump-range[0] intervals by the specified number of kbs. |
| --clump-range0 | With --clump and a 0-based region list file, report overlaps between clumps and regions. |
| --clump-test | Specify a different set of admissible --clump test-column value(s). |
| --clump-test-field | Specify a different --clump test field name search order. |
| --clump-unphased | Make --clump use unphased rather than phased r^2. |
| --clump{,-log10}-p1 | Adjust --clump index variant significance threshold. |
| --clump{,-log10}-p2 | Adjust --clump secondary significance threshold. |
| --condition | Add a single variant as a covariate for --glm. |
| --condition-list | Add variants named in a file as covariates for --glm. |
| --const-fid | Set FID = constant, IID = sample ID. |
| --covar | Specify covariate file. |
| --covar-col-nums | Specify covariate(s) in covariate file by column numbers/ranges. |
| --covar-name | Specify covariate(s) in covariate file by name. |
| --covar-quantile-normalize | Force numeric covariates to a N(0,1) distribution, preserving only the original rank orders. |
| --covar-variance-standardize | Linearly transform numeric covariates to mean-zero, variance 1. |
| --cow | Specify cow chromosome set. |
| --d | Specify non-dash delimiter for variant/covariate ranges. |
| --data | Oxford-format loader. |
| --debug | Flush new log entries to disk immediately, to simplify debugging of e.g. segfaults. |
| --delete-pmerge-result | Delete --pmerge[-list] result at end of run. |
| --dog | Specify dog chromosome set. |
| --dosage-erase-threshold | Erase dosages and keep only hardcalls when distance-from-hardcall ≤ the given threshold. |
| --double-id | Set both FID and IID to the input sample ID. |
| --dummy | Generate simple random dataset. |
| --error-on-freq-calc | Assert that the run does not calculate allele frequencies. |
| --exclude | Filter out all variants in the given file(s). |
| --exclude-if-info | Exclude variants which satisfy a comparison predicate on an INFO key. |
| --exclude-snp | Use ID to specify single variant to exclude, or center of window. |
| --exclude-snps | Use ID(s) to specify variant range(s) to exclude. |
| --export | Export genotype/dosage data to a non-PLINK 2 format. |
| --export-allele | Specify which allele counts to report with --export A/AD. |
| --extract | Filter out all variants not in the union of the provided file(s). |
| --extract-col-cond | Exclude all variants without a 2nd-column entry in the given file satisfying a condition. |
| --extract-col-cond-... | Specify --extract-col-cond condition. |
| --extract-if-info | Exclude variants which don't satisfy a comparison predicate on an INFO key. |
| --extract-intersect | Filter out all variants not in the intersection of the provided files. |
| --fa | Specify full name of reference FASTA file. |
| --fam | Specify full name of input .fam file. |
| --family | Use FIDs to define categories. |
| --family-missing-catname | Make --family treat the specified FID as missing. |
| --force-intersect | Allow multiple variant inclusion filters to be specified, and take the intersection. |
| --freq | Allele frequency report. |
| --from/--to | Use ID(s) to specify variant range to load. |
| --from-bp/--to-bp | Use physical bp position(s) to specify variant range to load. |
| --from-kb/--to-kb | Use physical kb position(s) to specify variant range to load. |
| --from-mb/--to-mb | Use physical mb position(s) to specify variant range to load. |
| --fst | Estimate Wright's Fst between all population pairs. |
| --gen | Specify full name of input Oxford-format .gen or .gen.gz file. |
| --geno | Filter out variants with many missing calls. |
| --geno-counts | Genotype count report. |
| --genotyping-rate | Print genotyping rate (was automatic in PLINK 1.x). |
| --glm | Linear/logistic/Firth-regression based association analysis. |
| --gwas-ssf | Reformat --glm output for GWAS Catalog. |
| --haps | Specify full name of input Oxford .haps file. |
| --hard-call-threshold | Adjust threshold beyond which a dosage is represented by a missing hardcall instead of the nearest one. |
| --hardy | Hardy-Weinberg equilibrium exact test p-value report. |
| --help | Look up command usage information. |
| --het | Inbreeding coefficient report. |
| --horse | Specify horse chromosome set. |
| --human | Specify human chromosome set. |
| --hwe | Filter out variants failing a Hardy-Weinberg equilibrium exact test. |
| --id-delim | Try to split sample IDs into FID/IID separated by a delimiter. |
| --idspace-to | Convert spaces in VCF/.bgen sample IDs to another character. |
| --iid-sid | Make --id-delim and --sample-diff treat 2-token sample IDs as IID-SID instead of FID-IID. |
| --import-dosage | Import PLINK 1 dosage file. |
| --import-dosage-certainty | Import missing call instead of dosage when largest genotype probability is less than given threshold. |
| --import-max-alleles | During VCF/BCF/BGEN import, filter out variants with more than the given # of alleles. |
| --impute-sex | Impute sexes from chrX/chrY genotype calls. |
| --indep | Use Variance Inflation Factors to generate a pruned marker list. |
| --indep-order | Set LD-pruning order. |
| --indep-pairphase | Use maximum likelihood phasing r^2 values to generate a pruned marker list. |
| --indep-pairwise | Use allele count pairwise correlations to generate a pruned marker list. |
| --indep-preferred | Make LD-pruning commands try to keep the variants listed in a file. |
| --indiv-sort | Select sort order for sample IDs in a new binary fileset. |
| --indv | Filter out all samples not matching the given ID. |
| --input-missing-genotype | Specify non-'.' genotype code to treat as missing in input files (default '0'). |
| --input-missing-phenotype | Specify integer phenotype code to treat as missing value (default -9). |
| --keep | Filter out all samples not named in a file. |
| --keep-autoconv | Keep autogenerated binary fileset when working with VCF/.bgen/other non-native data formats. |
| --keep-cat-names | Filter out all but the given categories (unless --keep-cats also specified). |
| --keep-cat-pheno | Specify which categorical phenotype --keep-cats/--keep-cat-names applies to. |
| --keep-cats | Filter out categories not named in a file. |
| --keep-col-match | Exclude samples without a 3rd-column entry in the given file matching a given string. |
| --keep-col-match-name | Make --keep-col-match match against the column with the given name, instead of the 3rd column. |
| --keep-col-match-num | Make --keep-col-match match against the given column number, instead of the 3rd column. |
| --keep-fam | Filter out all families not named in a file. |
| --keep-females | Exclude male and unknown-sex samples. |
| --keep-founders | Exclude nonfounder samples. |
| --keep-if | Exclude samples which don't satisfy a pheno/covar-based comparison predicate. |
| --keep-males | Exclude female and unknown-sex samples. |
| --keep-nonfounders | Exclude founder samples. |
| --keep-nosex | Exclude all known-sex samples. |
| --king-cutoff | Prune samples to remove close relations. |
| --king-cutoff-table | Prune samples to remove close relations, using precomputed .kin0 kinship values. |
| --king-table-filter | Specify minimum kinship coefficient for inclusion in --make-king-table report. |
| --king-table-require | Restrict --make-king-table run to sample pairs including at least one sample listed in a file. |
| --king-table-require-xor | Restrict --make-king-table run to sample pairs including exactly one sample listed in a file. |
| --king-table-subset | Restrict --make-king-table run to sample pairs listed in the given .kin0 file. |
| --lambda | Set genomic control lambda for --adjust[-file]. |
| --lax-bgen-import | Tolerate overstated variant count in .bgen header. |
| --lax-chrx-import | Disable warning/error for suspicious chrX import that seems to be missing sex information and/or --split-par. |
| --ld | Display LD statistics for a single pair of variants. |
| --ld-snp-list | Restrict the first variant in --r[2]-[un]phased pairs to those named in a file. |
| --ld-snp/--ld-snps | Restrict the first variant in --r[2]-[un]phased pairs to the given variant(s). |
| --ld-window | Set max variant-ct distance for --r[2]-[un]phased report inclusion. |
| --ld-window-cm | Set max centimorgan distance for --r[2]-[un]phased report inclusion. |
| --ld-window-kb | Change max kb distance for --r[2]-[un]phased report inclusion. |
| --ld-window-r2 | Change min r^2 threshold for --r[2]-[un]phased report inclusion. |
| --legend | Specify full name of input Oxford .legend file. |
| --loop-cats | Execute PLINK 2's main body once for each category. |
| --mac | Filter out variants with low minor allele count. |
| --mach-r2-filter | Filter out variants based on MaCH imputation quality metric. |
| --maf | Filter out variants with low MAF. |
| --maj-ref | Set major alleles to reference, like PLINK 1.x automatically did. |
| --make-bed | Generate a PLINK 1 .bed + .bim[.zst] + .fam binary fileset. |
| --make-bpgen | Generate a hybrid .pgen + .bim[.zst] + .fam binary fileset. |
| --make-founders | Set missing parental ID codes to 0. |
| --make-grm-bin | Export a GCTA 1.1+ triangular binary relationship matrix. |
| --make-grm-list | Export a GCTA relationship matrix (original format). |
| --make-just-bim | Generate a new PLINK 1 .bim file. |
| --make-just-fam | Generate a new PLINK 1 .fam file. |
| --make-just-psam | Generate a new PLINK 2 .psam file. |
| --make-just-pvar | Generate a new PLINK 2 .pvar file. |
| --make-king | KING-robust kinship estimator, matrix output. |
| --make-king-table | KING-robust kinship estimator, table output. |
| --make-pgen | Generate a PLINK 2 .pgen + .pvar[.zst] + .psam binary fileset. |
| --make-rel | Relationship/covariance matrix calculation. |
| --map | Specify full name of input .map file. |
| --max-alleles | Filter out variants with more than the given # of alleles. |
| --max-corr | Adjust --glm's predictor correlation limit (default 0.999). |
| --max-mac | Filter out variants with high minor allele count. |
| --max-maf | Filter out variants with high MAF. |
| --memory | Specify the size of the primary memory allocation request. |
| --merge-cm-mode | Adjust CM column conflict resolution behavior of --pmerge[-list]. |
| --merge-filter-mode | Adjust FILTER conflict resolution behavior of --pmerge[-list]. |
| --merge-info-mode | Adjust INFO conflict resolution behavior of --pmerge[-list]. |
| --merge-info-sort | Adjust INFO key sort order in --pmerge[-list] output. |
| --merge-max-allele-ct | Exclude variants with more alleles from --pmerge[-list] output. |
| --merge-mode | Adjust genotype/dosage conflict resolution behavior of --pmerge[-list]. |
| --merge-par | Merge PAR1/PAR2 with chrX. (Don't use this with "--export vcf".) |
| --merge-parents-mode | Adjust parental-ID conflict resolution behavior of --pmerge[-list]. |
| --merge-pheno-mode | Adjust phenotype conflict resolution behavior of --pmerge[-list]. |
| --merge-pheno-sort | Adjust phenotype column sort order in --pmerge[-list] output. |
| --merge-qual-mode | Adjust QUAL conflict resolution behavior of --pmerge[-list]. |
| --merge-sex-mode | Adjust sex conflict resolution behavior of --pmerge[-list]. |
| --merge-x | Merge PLINK 1.x XY pseudo-autosomal region with rest of chrX. |
| --merge-xheader-mode | Adjust .pvar header conflict resolution behavior of --pmerge[-list]. |
| --min-alleles | Filter out variants with fewer than given # of alleles. |
| --mind | Filter out samples with many missing calls. |
| --minimac3-r2-filter | Compute Minimac3 R^2s (inaccurate without phased dosages!), and filter variants on them. |
| --missing | Sample- and variant-based missing data reports. |
| --missing-catname | Set missing-categorical-phenotype string (case-sensitive, default 'NONE'). |
| --missing-code | Specify missing phenotype values in Oxford-formatted input. |
| --missing-var-code | Change unnamed variant code. |
| --mouse | Specify mouse chromosome set. |
| --multiallelics-already-joined | Prevent --pmerge[-list] from erroring out when an apparent 'split' multiallelic variant is seen. |
| --mwithin | Specify column to read --within category assignments from. |
| --native | Allow Intel MKL to use processor-dependent code paths. |
| --neg9-pheno-really-missing | Specify that -9 really represents a missing value when other values in [-8, -10] are present. |
| --new-id-max-allele-len | Set max # of chars --set-{all,missing}-var-ids can use from an allele name. |
| --no-categorical | Interpret non-numeric non-NA/nan phenotype/covariate values as missing numbers, instead of category names. |
| --no-fid | Specify that main input .fam/.ped file does not have a family ID column. |
| --no-id-header | Exclude header line when creating sample .id files. |
| --no-input-missing-phenotype | Treat phenotype=-9 as actually -9 instead of a missing code. |
| --no-parents | Specify that main input .fam/.ped file does not have parental ID columns. |
| --no-psam-pheno | Specify that phenotype data should not be loaded from the main input .psam/.fam file. |
| --no-sex | Specify that main input .fam/.ped file does not have a sex column. |
| --nonfounders | Include all samples in MAF/MAC/HWE test(s) and --freq report, instead of excluding nonfounders. |
| --normalize | Left-normalize all variants, using the --fa file. |
| --not-chr | Exclude all variants on the specified chromosomes. |
| --not-covar | Ignore the named covariates. |
| --not-pheno | Ignore the named phenotypes. |
| --out | Specify filename prefix to use for all output files. |
| --output-chr | Specify chromosome coding scheme in output files. |
| --output-min-p | Specify minimum p-value to write to reports. |
| --output-missing-genotype | Specify missing genotype code to use in PLINK-format output files. |
| --output-missing-phenotype | Specify missing phenotype code to use in PLINK-format output files. |
| --oxford-single-chr | Specify single-chromosome .gen/.bgen file. |
| --parallel | Split a computation into many parts and complete one of them. |
| --parameters | Exclude some covariates/interactions from --glm model. |
| --pca | Extract principal components. |
| --ped | Specify full name of input .ped file. |
| --pedmap | Import .ped + .map fileset. |
| --pfile | Make <prefix>.pgen + .pvar + .psam the main input fileset. |
| --pfilter | Only include variants with small p-values in association test reports. |
| --pgen | Specify full name of input .pgen file. |
| --pgen-diff | Compare overlapping samples and variants between two PLINK binary filesets. |
| --pgen-info | Print .pgen dimensions, and which data types are present. |
| --pgi | Specify full name of input .pgen.pgi file. |
| --pheno | Load phenotype data from the given file instead of the main input fileset. |
| --pheno-col-nums | Specify phenotype(s) in phenotype file by column numbers/ranges. |
| --pheno-inner-join | Restrict --pmerge[-list] to phenotypes present in every input filesets. |
| --pheno-name | Specify name of the column to load in --pheno file. |
| --pheno-quantile-normalize | Force quantitative phenotypes to a N(0,1) distribution, preserving only the original rank orders. |
| --pheno-svd | Generate summary phenotypes via SVD. |
| --pmerge | Merge a PLINK binary fileset with the main input fileset. |
| --pmerge-list | Merge multiple PLINK binary filesets. |
| --pmerge-list-dir | Specify base directory for --pmerge-list entries. |
| --pmerge-output-vzs | Compress the .pvar produced by --pmerge[-list]. |
| --polyploid-mode | Specify how polyploid data should be imported. |
| --psam | Specify full name of input .psam file. |
| --pvar | Specify full name of input .pvar file. |
| --q-score-range | Force --score[-list] to only consider e.g. SNPs with p-values in a given range. |
| --quantile-normalize | Force numeric covariates and quantitative phenotypes to a N(0,1) distribution, preserving only the original rank orders. |
| --r[2]-[un]phased | Write LD-statistic matrix/table to disk. |
| --randmem | Randomize contents of workspace memory immediately after reserving it. |
| --read-freq | Load allele frequencies from a --freq/--geno-counts file. |
| --recover-var-ids | Undo --set-all-var-ids, given the original .pvar/VCF/.bim file. |
| --real-ref-alleles | Specifies that .bim A2 alleles are based on a real reference genome. |
| --ref-allele | Scrape REF alleles from a VCF or a similar file. |
| --ref-from-fa | Set REF alleles from a reference FASTA. |
| --remove | Filter out all samples named in a file. |
| --remove-cat-names | Filter out all given categories. |
| --remove-cat-pheno | Specify which categorical phenotype --remove-cats/--remove-cat-names applies to. |
| --remove-cats | Filter out all categories named in a file. |
| --remove-fam | Filter out all families named in a file. |
| --remove-females | Exclude female samples. |
| --remove-if | Exclude samples which satisfy a pheno/covar-based comparison predicate. |
| --remove-males | Exclude male samples. |
| --remove-nosex | Exclude unknown-sex samples. |
| --require-covar | Remove samples with missing covariate values. |
| --require-info | Exclude variants which are missing one of the given INFO keys. |
| --require-no-info | Exclude variants which have any of the given INFO keys. |
| --require-pheno | Remove samples with missing phenotype value(s). |
| --rerun | Reuse flags (and parameters, except when explicitly replaced) from last run. |
| --rice | Specify rice chromosome set (assuming doubled haploids). |
| --rm-dup | Deduplicate same-ID variants. |
| --sample | Specify full name of input Oxford-format .sample file. |
| --sample-counts | Sample variant-count report (# of hom-ALT and het-ALT variants, etc.). |
| --sample-diff | Pairwise sample comparison. |
| --sample-inner-join | Restrict --pmerge[-list] to samples present in every input filesets. |
| --score-col-nums | Make --score[-list] process multiple coefficient columns from each input file. |
| --score[-list] | Compute sample scores, given allele coefficients. |
| --script | Read additional command-line parameters from a text file. |
| --seed | Specify seed(s) to initialize the pseudorandom number generator with. |
| --set-all-var-ids | Assign chromosome-and-position-based IDs to all variants. |
| --set-invalid-haploid-missing | Force heterozygous haploid genotypes to missing in --make-[b]pgen/--make-bed output. |
| --set-missing-var-ids | Assign chromosome-and-position-based IDs to unnamed variants. |
| --set-mixed-mt-missing | Force heterozygous MT genotypes to missing in --make-[b]pgen/--make-bed output. |
| --sheep | Specify sheep chromosome set. |
| --silent | Suppress all console output. |
| --snp | Use ID to specify single variant to load, or center of window. |
| --snps | Use ID(s) to specify variant range(s) to load. |
| --snps-only | Exclude all sites with multi-character allele codes. |
| --sort-vars | Sort variants by chromosome, then position, then ID. |
| --split-cat-pheno | Split categorical phenotypes into binary phenotypes. |
| --split-par | Split PAR1/PAR2 off of chrX. |
| --strict-extra-chr | Prohibit unrecognized chromosome codes, unless --allow-extra-chr is also present. |
| --strict-sid0 | When the main .psam/.fam file lacks a SID column, treat nonzero SIDs in other files as sample ID mismatches. |
| --tests | Perform the specified joint test on the --glm model. |
| --tfam | Specify full name of input .tfam file. |
| --tfile | Import .tped + .tfam fileset. |
| --thin | Randomly remove variants at a given rate. |
| --thin-count | Randomly remove variants until the given number remain. |
| --thin-indiv | Randomly remove samples at a given rate. |
| --thin-indiv-count | Randomly remove samples until the given number remain. |
| --threads | Specify the maximum permitted number of compute threads. |
| --tped | Specify full name of input .tped file. |
| --update-alleles | Update variant allele codes. |
| --update-chr | Update variant chromosome codes. |
| --update-cm | Update variant centimorgan positions. |
| --update-ids | Update sample IDs. |
| --update-map | Update variant bp positions. |
| --update-name | Update variant IDs. |
| --update-parents | Update parental IDs. |
| --update-sex | Update sexes. |
| --validate | Scan .pgen for malformed records. |
| --var-filter | Skip variants failing one or more filters tracked by the FILTER field. |
| --var-id-multi | Specify alternative --set-{all,missing}-var-ids template for multiallelic variants. |
| --var-id-multi-nonsnp | Specify alternative --set-{all,missing}-var-ids template for non-SNP multiallelic variants. |
| --var-min-qual | Skip variants with missing or low QUAL values. |
| --variance-standardize | Linearly transform quantitative phenotypes and numeric covariates to mean-0, variance 1. |
| --variant-inner-join | Restrict --pmerge[-list] to variants present in every input filesets. |
| --variant-score | Compare variant scores, given sample coefficients. |
| --vcf | Load text VCF file (can be gzipped). |
| --vcf-allow-no-nonvar | Disable warning/error for suspicious single-sample VCF with no 0, 0/0, or 0|0 GT values. |
| --vcf-half-call | Specify how '0/.' and similar nonstandard VCF GT values should be handled. |
| --vcf-max-dp | No-call genotypes when DP is present and above the given threshold. |
| --vcf-min-dp | No-call genotypes when DP is present and below the given threshold. |
| --vcf-min-gq | No-call genotypes when GQ is present and below the given threshold. |
| --vcf-ref-n-missing | Import VCF 'N' REF alleles as missing alleles. |
| --vcf-require-gt | Skip variants with no GT field. |
| --version | Print version information and exit. |
| --vif | Specify VIF threshold for --glm multicollinearity check. |
| --vscore-col-nums | Make --variant-score only process the specified coefficient columns in the input file. |
| --warning-errcode | Return a nonzero error code to the OS when a run completes with warning(s). |
| --window | Specify total size of variant window to keep/exclude. Requires --snp or --exclude-snp. |
| --within | Construct categorical covariate from PLINK 1 cluster file. |
| --write-covar | Create a pruned covariate file. |
| --write-samples | Export a list of sample IDs. |
| --write-snplist | Export a list of variant IDs. |
| --xchr-model | Specify chrX male dosage encoding for --glm/--condition[-list]/--score[-list]/--vscore. |
| --y-nosex-missing-stats | Include unknown-sex samples in chrY missingness-rate and het-haploid stats. |
| --zst-decompress | Decompress .zst file. |
| --zst-level | Adjust Zstd compression level (1-22, default 3). |