scran::FitVarianceTrend Class Reference

Fit a mean-variance trend to log-count data. More...

#include <FitVarianceTrend.hpp>

Classes
struct	Defaults
	Parameter defaults for trend fitting. More...
struct	Results
	Results of the trend fit. More...

Public Member Functions
FitVarianceTrend &	set_span (double s=Defaults::span)
FitVarianceTrend &	set_minimum_mean (double m=Defaults::minimum_mean)
FitVarianceTrend &	set_filter (bool f=Defaults::filter)
FitVarianceTrend &	set_transform (bool t=Defaults::transform)
FitVarianceTrend &	set_use_fixed_width (bool u=Defaults::use_fixed_width)
FitVarianceTrend &	set_fixed_width (double f=Defaults::fixed_width)
FitVarianceTrend &	set_minimum_window_count (int c=Defaults::minimum_window_count)
void	run (size_t n, const double *mean, const double *variance, double *fitted, double *residuals) const
Results	run (size_t n, const double *mean, const double *variance) const

Detailed Description

Fit a mean-variance trend to log-count data.

Fit a trend to the per-feature variances against the means, both of which are computed from log-normalized expression data. We use a LOWESS smoother in several steps:

1. Filter out low-abundance genes, to ensure the span of the smoother is not skewed by many low-abundance genes.
2. Take the quarter-root of the variances, to squeeze the trend towards 1. This makes the trend more "linear" to improve the performance of the LOWESS smoother; it also reduces the chance of obtaining negative fitted values.
3. Apply the LOWESS smoother to the quarter-root variances. This is done using the implementation in the WeightedLowess library.
4. Reverse the quarter-root transformation to obtain the fitted values for all non-low-abundance genes.
5. Extrapolate linearly from the left-most fitted value to the origin to obtain fitted values for the previously filtered genes. This is empirically justified by the observation that mean-variance trends of log-expression data are linear at very low abundances.

Member Function Documentation

set_span()

FitVarianceTrend & scran::FitVarianceTrend::set_span ( double  s = Defaults::span )

inline

Set the span for the LOWESS smoother as a proportion of the total number of points. This is only used if set_use_fixed_width() is set to false.

Parameters

s	Span for the smoother.

Returns

A reference to this FitVarianceTrend object.

set_minimum_mean()

FitVarianceTrend & scran::FitVarianceTrend::set_minimum_mean ( double  m = Defaults::minimum_mean )

inline

Set the minimum mean log-expression, to use for filtering out low-abundance genes.

Parameters

m	Minimum value for the mean log-expression.

Returns

A reference to this FitVarianceTrend object.

set_filter()

FitVarianceTrend & scran::FitVarianceTrend::set_filter ( bool  f = Defaults::filter )

inline

Should any filtering be performed by log-expression? This may need to be disabled if FitVarianceTrend is not being used on statistics computed from log-expression values.

Parameters

f	Whether filtering should be performed.

Returns

A reference to this FitVarianceTrend object.

set_transform()

FitVarianceTrend & scran::FitVarianceTrend::set_transform ( bool  t = Defaults::transform )

inline

Should any transformation of the variances be performed prior to LOWESS smoothing. This may need to be disabled if FitVarianceTrend is not being used on statistics computed from log-expression values.

Parameters

t	Whether transformation should be performed.

Returns

A reference to this FitVarianceTrend object.

set_use_fixed_width()

FitVarianceTrend & scran::FitVarianceTrend::set_use_fixed_width ( bool  u = Defaults::use_fixed_width )

inline

Should a fixed-width constraint be applied to the LOWESS smoother? This forces each window to be a minimum width (see fixed_width()) and avoids problems with large differences in density. For example, the default smoother performs poorly at high abundances where there are few genes.

Parameters

u	Whether to apply fixed-width constraints.

Returns

A reference to this FitVarianceTrend object.

set_fixed_width()

FitVarianceTrend & scran::FitVarianceTrend::set_fixed_width ( double  f = Defaults::fixed_width )

inline

Define the width of the window to use when set_use_fixed_width() is set to true. This should be relative to the range of mean values in run(); the default value is chosen based on the typical range in single-cell RNA-seq data.

Parameters

f	Fixed width of the window.

Returns

A reference to this FitVarianceTrend object.

set_minimum_window_count()

FitVarianceTrend & scran::FitVarianceTrend::set_minimum_window_count ( int  c = Defaults::minimum_window_count )

inline

Define the minimum number of observations in each window when set_use_fixed_width() is set to true. This ensures that each window contains at least a given number of observations; if it does not, it is extended using the standard LOWESS logic until the minimum number is achieved.

Parameters

c	Minimum number of observations in the window.

Returns

A reference to this FitVarianceTrend object.

run() [1/2]

void scran::FitVarianceTrend::run ( size_t  n, const double *  mean, const double *  variance, double *  fitted, double *  residuals  ) const

inline

Run the trend fitting on the means and variances across all features. This returns the fitted value and residual from the trend for each feature.

Parameters

	n	Number of features.
[in]	mean	Pointer to an array of length n, containing the means for all features.
[in]	variance	Pointer to an array of length n, containing the variances for all features.
[out]	fitted	Pointer to an array of length n, to store the fitted values.
[out]	residuals	Pointer to an array of length n, to store the residuals.

run() [2/2]

Results scran::FitVarianceTrend::run ( size_t  n, const double *  mean, const double *  variance  ) const

inline

Run the trend fitting on the means and variances across all features.

Parameters

	n	Number of features.
[in]	mean	Pointer to an array of length n, containing the means for all features.
[in]	variance	Pointer to an array of length n, containing the variances for all features.

Returns

A Results object containing the fitted values and residuals of the trend.

---

The documentation for this class was generated from the following file:

• scran/feature_selection/FitVarianceTrend.hpp