Split the bins that are too large

optimize_bin_max() splits every bin that holds more than max_allowed points into the number of bins that method and drop give for its points (estBinSize()), of equal width over their range; the first new bin starts at the lower bound of the bin and the last ends at its upper bound, so the new bins cover it without gaps. The other bins are kept.

Usage

optimize_bin_max(
  bin_table,
  max_allowed,
  verbose = TRUE,
  time_vector,
  lbound,
  ubound,
  bin.size,
  method,
  drop
)

Arguments

bin_table

A data frame of bins, one row per bin, with the columns lbound, ubound and bin.size.

max_allowed

The largest number of points of a bin that is kept.

verbose

Whether to print a message for each bin.

time_vector

The pseudotime of the cells of the path.

lbound, ubound

The names of the columns of the lower and upper bounds.

bin.size

The name of the column of the number of points.

method

The rule that gives the number of new bins.

drop

The factor that multiplies that number (drop_fac).

Value

A data frame with the columns of bin_table, one row per bin, where each split bin is replaced by its parts.

Details

sc.squeeze() passes as max_allowed the maximum of Srivastava et al. (2024), Section 2.2: the mean plus the standard deviation of the sizes of the bins of the path, set once from the bins of the rule. "Bins larger than the maximum size are divided", and "The process is repeated until all bins conform to the established size criteria": sc.squeeze() calls optimize_bin_max() again, with the same max_allowed, until no bin is larger.

See also

Other binning helpers: calc_bin_size(), estBinSize(), extract_interval(), pb_counts()