asinh operator

Description

asinh operator performs an inverse hyperbolic sine on values. Also denoted as arcsinh, arsinh, or argsinh.

This is a Docker-based operator running in the ghcr.io/tercen/flowvs container, which includes the flowVS package from Bioconductor for automatic cofactor estimation.

Usage

Input projection	.
y-axis	numeric, values required to be transformed by the asinh operator
row	channels, and scale values (optional, required for manual/auto methods)
col	event, for example

Input parameters	.
method	string, transformation method: 'fixed' (default), 'manual', or 'auto'
scale	numeric, the scaling factor to use before the asinh transformation (only used when method is 'fixed'), default is 5

Output relations	.
asinh	numeric, output transformation per row and col.

Details

Values are scaled first and then asinh transformation is performed. One data point per cell is required as input.

asinh(value/scale)

Methods

fixed (default) Uses a global scale parameter for all channels. A scale of 5 is recommended for CyTOF measurements and 150 for flow cytometry measurements.

manual Reads per-channel scale values from the row factor. Requires a second factor in the row dimension that provides the scaling value for each channel.

auto Automatically estimates optimal cofactors per channel using the flowVS variance stabilization algorithm from Bioconductor. This method finds the cofactor that minimizes variance heterogeneity across cell populations identified in each channel.

The algorithm:

1. Identifies cell populations as density peaks using kernel density estimation
2. For each candidate cofactor, transforms data with asinh(x/cofactor)
3. Calculates Bartlett's statistic to measure variance homogeneity across populations
4. Uses optimization to find the cofactor that minimizes Bartlett's statistic

Requirements for auto method:

• Channel names must be provided in the row projection
• At least 100 data points per channel for reliable estimation
• Multiple samples recommended for robust cofactor estimation

References

• See the base::asinh function of the R package for more information.
• Azad A, Rajwa B, Pothen A (2016). "flowVS: Channel-Specific Variance Stabilization in Flow Cytometry." BMC Bioinformatics, 17:291. doi:10.1186/s12859-016-1083-9