0.1 | Filter variants by minor allele frequency (0-1). Variants with MAF below this threshold will be hiddenNumber.MAX_SAFE_INTEGER | Maximum length of variants to include (in bp)100 | Height of the zone for connecting lines at the top'r2' | LD metric to compute: 'r2' (squared correlation) or 'dprime' (normalized D) |
| **showLegend**false _promotable_ | Whether to show the legend. Unset (the default) follows the session-wide default for this display type, falling back to off; an explicit true/false customizes the track. |
| **showLDTriangle**true | Whether to show the LD triangle heatmap |
| **squashToHeight**false | When true, squash the LD triangle to fit the display height0 | HWE filter p-value threshold (variants with HWE p < this are excluded). Set to 0 to disable HWE filtering0 | Call rate filter threshold (0-1). Variants with fewer than this proportion of non-missing genotypes are excluded. Set to 0 to disable.0 | Maximum separation, in variants, between the two SNPs of a computed pair. Pairs further apart are not computed and not drawn, which turns the matrix from n²/2 cells into n·k and so makes the cost linear in the variant count rather than quadratic. This is plink's `--ld-window`, and like it the window is the way to look at a large region at all: the full triangle for 50,000 variants is 1.25e9 cells, which no GPU will allocate. Set to 0 for the full triangle.true | Whether to show vertical guides at the connected genome positions on hoverfalse | Whether to show variant labels above the tick marks6 | Height of the vertical tick marks at the genomic positionfalse | When true, draw cells sized according to genomic distance between SNPs rather than uniform squaresfalse | When true, show signed LD values (-1 to 1) instead of absolute values (0 to 1). For R², this shows R (correlation) instead. For D', this preserves the sign.'auto' | Which LD estimator to compute. 'auto' picks haplotypic LD for a phased callset and the Weir composite estimate for an unphased one, which is the most precise statistic each file can support. 'composite' forces the composite estimate even on phased data — the two are different statistics that coincide only under Hardy-Weinberg, so forcing it is how a phased panel is made comparable to an unphased cohort, or to plink `--r2` output. 'phased' is a preference rather than an instruction: unphased data carries no gametes to count, so it declines to composite, and `method` on the result reports what actually ran.400 | Starting height in pixels for the LD triangle, excluding the lineZoneHeight band; drag-resizable |
| Inherited from [BaseLinearDisplay](../baselineardisplay) | 5 slots |
| **maxFeatureScreenDensity**1 | maximum features per pixel before showing a "too many features" message[] | config jexlFilters are deferred evaluated so they are prepended with jexl at runtime rather than being stored with jexl in the config |
| **fetchSizeLimit**1_000_000 | maximum data to attempt to download for a given track, used if adapter doesn't specify onefalse | Declarative equivalent of the "Force load" button on the "too much data" banner: when true the display always renders, however large the region or dense the features. Off by default (the gate guards against huge downloads). Set it on a view no one can interact with — an embedded / notebook view, or a screenshot — where the region is known and you want it drawn without a click.