Reformatting strategies

Two main strategies are available:

1. Base-wise Statistics

Each signal segment aligned to a base of interest is summarized into statistics that represent its characteristics.

Statistic	Description
mean	Mean signal intensity
median	Median signal intensity
std	Standard deviation of signal intensity
dwell	Dwell time (number of signal samples assigned to the base)
signal-to-noise	Signal-to-noise ratio (mean / std)

This is the default strategy, but can be enabled explicitly with --strategy stats. By default, mean, std, and dwell are calculated. You can specify a custom subset via:

--stats <STAT-A> <STAT-B> ...

2. Interpolation into uniform shapes

Instead of condensing each segment into statistics, the signal is reshaped into a fixed number of samples per base using linear interpolation. This allows direct comparison or machine-learning-based analysis across bases.

To select this strategy, set --strategy interpolate. The number of samples per base can be tuned using --target-size <NUM-SAMPLES> (default: 30)

Upsampling or downsampling is applied as needed.