Fit Chromatin Accessibility-Sparsity Polynomial Curve (simATAC)

Fits a polynomial curve relating peak/bin mean accessibility to non-zero cell proportion (NZP / detection frequency) as modeled in the simATAC framework (Navidi et al., Genome Biology 2021): NZP = c0 + c1 * mean + c2 * mean^2.

Usage

calc_accessibility_sparsity_curve(data, poly_degree = 2)

Arguments

data

Count matrix (features/peaks x cells) or data.frame with 'mean' and 'nzp'.

poly_degree

Degree of polynomial (default: 2 for quadratic curve).

Value

A list with estimated coefficients (c0, c1, c2), R-squared, Spearman/Pearson correlations, and model fit summary.

Examples

data(example_scrna, package = "scSimEval")
calc_accessibility_sparsity_curve(example_scrna$ref)
#> $c0
#> [1] 0.5117571
#> 
#> $c1
#> [1] 0.1032392
#> 
#> $c2
#> [1] -0.006616585
#> 
#> $r_squared
#> [1] 0.7560448
#> 
#> $spearman_cor
#> [1] 0.8167929
#> 
#> $pearson_cor
#> [1] 0.8428295
#> 
#> $poly_degree
#> [1] 2
#> 
#> $model
#> 
#> Call:
#> stats::lm(formula = y ~ stats::poly(x, degree = poly_degree, 
#>     raw = TRUE), data = df)
#> 
#> Coefficients:
#>                                       (Intercept)  
#>                                          0.511757  
#> stats::poly(x, degree = poly_degree, raw = TRUE)1  
#>                                          0.103239  
#> stats::poly(x, degree = poly_degree, raw = TRUE)2  
#>                                         -0.006617  
#> 
#>