Bug fix to lag_time where it had been returning incorrect values if trans_y = "log" and the y values provided were not already normalized (i.e. did not have the blank value subtracted). lag_time now takes a blank argument, which is required when trans_y = "log"
New vignette section on subtracting blanks
The join_designs argument in import_blockdesigns has been renamed to join_as_cols for clarity
Citation and load message now point to the BMC Bioinformatics paper for the package: Blazanin, M. gcplyr: an R package for microbial growth curve data analysis. BMC Bioinformatics 25, 232 (2024). https://doi.org/10.1186/s12859-024-05817-3
New functions centroid, centroid_x, centroid_y, and centroid_both calculate the center-of-mass centroid of the area under the curve, which can be used as a metric of microbial growth
import_blockdesigns now separates designs that were joined as rows when the sep argument is specified
separate_tidy now prints a message with inferred column names
Various documentation and vignette improvements
Bug fix to lag_time where it had been incorrectly calculating y0 as the minimum y value where deriv was not NA. It now calculates the minimum y value regardless of the deriv value
Bug fix to lag_time where NA was returned when there was only one non-NA data point, it now returns the x value of the single data point
When lag_time returns NA, it now always returns a numeric NA
Bug fix for import_blockdesigns where it had not been separating designs if they were read from only a single file
Bug fix for import_blockdesigns where designs were not being joined as columns if they were read from a single file
Bug fix for import_blockdesigns where arguments passed via ... were previously not being passed to separate_tidy
import_blockdesigns has a new argument join_designs which controls whether multiple imported designs should be handled as referring to the same plate (and so joined as columns), or as different plates (and so joined as rows)
import_blockdesigns now has an option to keep the block_names column in the output
solve_linear now correctly handles the case where x1 == x2 == x3
New arguments window_width_frac and window_width_n_frac in window-using functions (including find_local_extrema and related functions, calc_deriv, moving_average, and moving_median). These arguments allow the width of the window to be specified as a fraction of the total range of data or total number of data points, respectively
moving_average and moving_median have new arguments, x and y, that can be used instead of formula and data
A new function predict_interpolation works like stats::predict by linearly interpolating from existing data. This function can be used as the predict function for the output of moving_average and moving_median
New functions train_smooth_data and makemethod_train_smooth_data enable users to test different tuning parameter values for smooth_data via caret::train
Many functions now have warn_* arguments, allowing users to silence specific warnings
Bug fix in auc where warning was being incorrectly issued when xlim was inside the range of all x values but not inside the range of x values after NA x and y values had been removed
moving_average and moving_median now warn when not all x values are unique
Improved warnings in calc_deriv when window is not wide enough
lag_time now warns when the estimated lag time is less than the minimum non-NA value of x
Some internal functions were renamed for consistent style and more clear error messages
Various documentation and vignette improvements
read_ and import_ functions can now use any of the utils::read.table file types, including csv, csv2, delim, and delim2
The extension argument of read_ and import_ functions has been deprecated, the new argument name is filetype
The block_name_header argument in read_blocks has been deprecated, the new argument name is block_names_header
The names_to_col argument in read_wides and read_tidys has been deprecated, the new argument name is run_names_header
read_ and import_ functions have new arguments (_dot, _path, and _ext), which users can set to determine how file names/paths are saved in the resulting object
smooth_data has a new smoothing method: smooth.spline. This method uses stats::smooth.spline via a gcplyr wrapper function gc_smooth.spline
The area under the curve function auc now has a subset argument
There is a new data object included in the package: example_design_tidy. This object is tidy-shaped and contains the plate layout design for the example_widedata and example_widedata_noiseless datasets that have been included with gcplyr
find_local_extrema and find_threshold_crosses now have consistent handling of the subset argument, including treating NA's in subset as FALSE with a warning
Bug fix where reading files failed if extension was not specified by user and extension inferred from the filename was not .xls, .xlsx, .csv, or .tbl
Bug fix for extr_val where na.rm = TRUE was not working correctly
Improved error message when user specifies argument with length 0
Various backend improvements, including renaming internal functions for consistent style and renaming internal objects to be more unique for less ambiguous error messages
Various vignette improvements
read_blocks now returns blocks whose row.names are always a character vector
merge_dfs can now do inner, left, and right joins (in addition to the original capacity to do full joins)
Strings to be coerced into NA can now be specified to separate_tidy
The directory to write files to can now be specified to write_blocks
In read_blocks the metadata field name where block names are stored can now be specified via the block_name_header argument
There is a new vignette demonstrating how to work with multiple plates at once
Examples throughout all vignettes are now created with a make_example function for streamlined readability
read_blocks, read_wides, read_tidys, and make_design now provide better error messages when inputs are out of range
merge_dfs now provides a message with the number of dropped rows when drop = TRUE
Fix to bug in read_blocks where arguments for read.csv could not be passed via ...
Fix to bug where merge_dfs would not drop incomplete cases unless both x and y were specified
Vignettes are now numbered so they are sorted correctly on CRAN
Various minor documentation improvements
Various backend improvements
The "Dealing with noise" vignette was rewritten to be much more concise
Other minor fixes for CRAN checks
Several fixes to bugs where calc_deriv had been returning incorrect values when trans_y = 'log' and either blank was specified or fitting was not used (i.e. window_width and window_width_n were both NULL)
Bug fixes and better warning messages in lag_time, especially relating to negative and infinite y-values following log-transformation
New function: solve_linear, which enables easy calculation of the slope between two points, or finding an additional point on a known line
New functions: which_min_gc, which_max_gc, min_gc, max_gc, and extr_val. These serve as versions of the base functions which.min, which.max, min, max, and [] (respectively) that have better defaults and handling of edge cases (often related to NA values) for growth curve analyses with dplyr::summarize
The auc function has two new arguments. 'blank' allows setting of a blank value, and 'neg.rm' gives users a choice in how to handle below-zero values
Various documentation improvements in both the manual and vignettes
There are planned changes for smooth_data with return_fitobject = TRUE: in a forthcoming version (the first version released after Sep 1, 2023), return_fitobject = TRUE will return a list of the object created by the underlying method directly. The current behavior modifies the generated object so that 'fitted' is always the first element and 'residuals' is always the second element
mdp() is a new function, it is simply a shorthand alias for make_designpattern()
Column numbers and row numbers in read_* and import_* functions can now be specified using a mix of base-10 numbers and base-26 Excel-style letters
separate_tidy now coerces strings of "NA" into NA values by default
Bug fix in make_design where NA values were being returned as "NA" (a string) instead of NA (missing value indicator) when output_format = 'tidy' and there were multiple design columns
Bug fix for make_design where pattern could not be just "1"
Bug fix where an error arose in calc_deriv when trans_y = 'log' was used with y values at or below 0. calc_deriv now handles the resulting NA/NaN values and raises a warning
Bug fixes in first_maxima and first_minima
Added warning when using make_design and it's likely that setting a custom lookup_tbl_start was forgotten
Improved documentation for make_designpattern
New function lag_time to calculate lag time
New function doubling_time to convert per-capita growth rate into equivalent doubling time
smooth_data and calc_deriv now pass a warning when they are used outside of dplyr::mutate or on ungrouped data
make_design has better warning messages for common mistakes
Major edits to most of the vignettes
Various documentation improvements
The function first_peak has been renamed to first_maxima. first_peak will continue to function normally for a long time but with a warning to use first_maxima instead.
New functions have been added that are shortcuts for common use cases: first_minima and first_above
Bug fixes for find_threshold_crossings related to return_endpoints = TRUE
Bug fixes for find_threshold_crossings and find_local_extrema when input values are NA
Style and content improvements in vignettes
Help pages have been improved
The default behavior of write_blocks has changed: users are now required to specify the file argument to be either NULL, a file name, or a vector of file names. For the old default naming, specify file = NULL. This change is required for CRAN compatibility.
Minor changes to citation
Tweaked documentation for CRAN compatibility
Readability improvements to "Dealing with noise" vignette
Minor tweaks to several vignettes to reduce build time for CRAN compatibility
Improved clarity in vignettes, especially in the calculating derivatives and analyzing data pages
New vignette on dealing with noise in data
Small tweaks to other documentation
There is a new citation. Run citation("gcplyr") to see the new version
smooth_data methods moving-average and moving-median now accept a new smoothing parameter: window_width
find_local_extrema now has arguments window_width, window_width_n, and window_height (for naming consistency with smooth_data and calc_deriv arguments). Arguments width_limit, width_limit_n, and height_limit have been deprecated.
calc_deriv can now calculate derivatives using a linear fit over multiple data points determined by arguments window_width and/or window_width_n. For per-capita derivatives, y-values can be fit as-supplied and divided by the mid-point y-value, or can be fit after log-transformation
The gcplyr-workflow vignette was split into multiple smaller vignettes
A new data.frame is included with gcplyr: example_data_noiseless. This data is the same as example_data but does not include any of the simulated noise present in example_data.
Some small numerical changes to example_data values occurred on re-generation of example_data.
Packages mgcv and readxl are now Suggests (previously they were Imports), with errors thrown when they are not installed but are required for gcplyr functionality
find_local_extrema is now much faster
A new vignette section on running statistics
In the vignette, use of dplyr::mutate in the smoothing and derivatives sections
This is the first public release of gcplyr as an Open Beta.
Documentation should be complete, although there are planned additions to the vignette that have not been completed and included here.
Functions and arguments should largely be stabilized, although some small changes may occur going forward following user feedback.
Internal tests are mostly complete, although additional edge cases may be added as bugs are reported.