How Do I Install An Old Version of PhotoGEA?
All PhotoGEA releases are tagged on GitHub, and the tag
names are formatted as vX.Y.Z, where X.Y.Z is
the version number. Because of this,
remotes::install_github can be used to install old versions
from within R, supplying the appropriate tag name as the
ref input argument. For example, version 1.0.0
can be installed as follows:
remotes::install_github('eloch216/PhotoGEA', ref = 'v1.0.0')Note that this command requires the remotes package,
which can be installed from within R by typing
install.packages('remotes').
How Do I Install The Development Version of PhotoGEA?
The development version of PhotoGEA contains the latest
changes, although it may be a “work in progress” and subject to suddent
changes without any warning. It can always be found on the
unreleased branch on GitHub, and the package can be
installed from this branch as follows:
remotes::install_github('eloch216/PhotoGEA', ref = 'unreleased')Note that this command requires the remotes package,
which can be installed from within R by typing
install.packages('remotes').
What Should I Do When check_response_curve_data Fails?
When beginning to work with a new data set, it is fairly common to
encounter problems that cause check_response_curve_data to
fail. The most common issues are:
The manual log button was accidentally pressed while measuring, causing a response curve to have an extra point.
One or more “User Constant” was not updated before starting a to measure a new curve, causing two curves to have the same identifying metadata.
A curve was started, but needed to be ended early, resulting in a curve with fewer points than expected.
These are exactly the kinds of issues that
check_response_curve_data is designed to detect. It may
feel irritating to encounter these errors, but it’s better to know about
issues with extra log points or user constants early in your analysis,
before they cause additional downstream issues.
In each of these cases, it is often simplest to fix the issues by manually editing the log files in Excel. This is an easy way to delete extra rows or alter the values of user constants. When altering log files, it is always a good idea to keep an “original” version in case any of the changes need to be reverted.
Sometimes check_response_curve_data fails for a
different reason – for example, perhaps different setpoint sequences
were intentionally used while measuring the curves, or perhaps you have
already cleaned up your data in Excel. In these cases, please see the
“What If My Response Curves Have Different Numbers of Points?” and “What
If I Clean My Data In Excel?” sections of this article.
What If My Response Curves Have Different Numbers of Points?
The analysis guides use data sets where each response curve has the
same number of points, making it easy to apply
check_response_curve_data and
organize_response_curve_data. However, data sets may
sometimes have curves measured using different sequences of
setpoints.
To deal with this situation, one strategy is to split the full set into groups that are expected to use the same sequences, then separately check and organize each group, and finally recombine all the groups back together. The code snippets below show two examples of how this could be accomplished.
Splitting response curves according to the number of points in each curve
The following code would replace the calls to
check_response_curve_data and
organize_response_curve_data in the user guides. This code
was originally written for a set of A-Ci curves that used different
numbers of recovery points. The user wished to keep the final recovery
point to use for subsequent analysis.
The curves used the following sequences of CO2_r
setpoint values:
16 points: 400, 300, 200, 120, 70, 30, 10, 400, 400, 400, 600, 800, 1200, 1500, 1800, 400
18 points: 400, 300, 200, 120, 70, 30, 10, 400, 400, 400, 400, 400, 600, 800, 1200, 1500, 1800, 400
19 points: 400, 300, 200, 120, 70, 30, 10, 400, 400, 400, 400, 400, 400, 600, 800, 1200, 1500, 1800, 400
# Add a new column called `curve_npts` that stores the number of points in each
# response curve
licor_data <- do.call(rbind, by(licor_data, licor_data[, 'curve_identifier'], function(x) {
x[, 'curve_npts'] <- nrow(x)
x
}))
# Choose points to remove, depending on how many points are in the curve
pts_to_remove <- list(
'16' = c(1, 8:9, 16),
'18' = c(1, 8:11, 18),
'19' = c(1, 8:12, 19)
)
# Check and process each group of curves depending on how many points are in the
# curve
licor_exdf_list_processed <- by(licor_data, licor_data[, 'curve_npts'], function(x) {
# Get the number of points in these curves
npts <- x[1, 'curve_npts']
# Make sure info is specified for this group of curves
if (!as.character(npts) %in% names(pts_to_remove)) {
stop('Points to remove were not specified for npts = `', npts, '`')
}
# Make sure the data meets basic requirements
check_response_curve_data(x, 'curve_identifier', npts, 'CO2_r_sp')
# Remove points with duplicated `CO2_r_sp` values and order by `Ci`
organize_response_curve_data(
x,
'curve_identifier',
pts_to_remove[[as.character(npts)]],
'Ci'
)
})
# Use `rbind` to recombine all the data
licor_data <- do.call(rbind, licor_exdf_list_processed)Splitting the response curves according to the date they were measured and the number of points in each curve
This code would replace the calls to
check_response_curve_data and
organize_response_curve_data in the user guides. This code
was originally written for a set of A-Ci curves that used different
sequences of CO2_r setpoints on different days.
The curves used the following sequences of CO2_r
setpoint values:
2023-03-21: 18 points: 400, 300, 200, 150, 100, 75, 50, 40, 30, 20, 10, 400, 400, 600, 800, 1000, 1200, 1500
2023-03-23: 19 points: 400, 300, 200, 150, 100, 75, 50, 40, 30, 20, 10, 400, 400, 500, 600, 800, 1000, 1200, 1500
2023-03-24: 19 points: 400, 300, 200, 150, 100, 75, 50, 40, 30, 20, 400, 400, 450, 500, 600, 800, 1000, 1200, 1500
# Add a new column called `curve_npts` that stores the number of points in each
# response curve
licor_data <- do.call(rbind, by(licor_data, licor_data[, 'curve_identifier'], function(x) {
x[, 'curve_npts'] <- nrow(x)
x
}))
# Add a new column called `date_ymd` that stores the date formatted as
# YYYY-MM-DD
licor_data[, 'date_ymd'] <- paste(
substring(licor_data[, 'date'], 1, 4),
substring(licor_data[, 'date'], 5, 6),
substring(licor_data[, 'date'], 7, 8),
sep = '-'
)
# Add a new column called `date_ymd_npts` that combines the date and the number
# of points
licor_data[, 'date_ymd_npts'] <-
paste(licor_data[, 'date_ymd'], licor_data[, 'curve_npts'], sep = ' - ')
# Choose points to remove, depending on the date the curve was measured and the
# number of points it contains
pts_to_remove <- list(
'2023-03-21 - 18' = c(12, 13),
'2023-03-23 - 19' = c(12, 13),
'2023-03-24 - 19' = c(11, 12)
)
# Check and process each group of curves depending on the date and the number of
# points
licor_exdf_list_processed <- by(licor_data, licor_data[, 'date_ymd_npts'], function(x) {
# Get the date and number of points in these curves
date_ymd_npts <- x[1, 'date_ymd_npts']
npts <- x[1, 'curve_npts']
# Make sure info is specified for this group of curves
if (!date_ymd_npts %in% names(pts_to_remove)) {
stop('Points to remove were not specified for date_ymd_npts = `', date_ymd_npts, '`')
}
# Make sure the data meets basic requirements
check_response_curve_data(x, 'curve_identifier', npts, 'CO2_r_sp')
# Remove points with duplicated `CO2_r_sp` values and order by `Ci`
organize_response_curve_data(
x,
'curve_identifier',
pts_to_remove[[date_ymd_npts]],
'Ci'
)
})
# Use `rbind` to recombine all the data
licor_data <- do.call(rbind, licor_exdf_list_processed)What If I Clean My Data In Excel?
In the user guides, the organize_response_curve_data and
remove_points functions are used to remove recovery points
and other points from sets of response curves. However, it is also
possible to remove points in Excel before reading log files into R. In
this case, it is necessary to make a few small alterations to the code
used in the user guides.
One consideration is that after cleaning the curves in Excel, it is
likely that not all curves have the same number of points or follow the
same sequence of setpoint values. Because of this, the checks in
check_response_curve_data are likely to fail. In this case,
we recommend setting expected_npts to 0 (the default value)
and error_on_failure to FALSE when calling
check_response_curve_data. This will provide potentially
useful information about the number of points in each curve, but won’t
throw an error that would cause a script to stop running.
Another consideration is that
organize_response_curve_data is not needed to remove any
points from the curves. Yet, its other features, such as reordering the
points and calculating average values, are still useful. In this case,
we recommend setting measurement_numbers_to_remove to
c() and leaving the other arguments as-is.
Putting this all together would produce something like the following
code, which would replace the regular calls to
check_response_curve_data and
organize_response_curve_data from the user guides:
# Print info about the number of points in each curve to make sure
# `curve_identifier` is able to properly split the set into individual curves
check_response_curve_data(licor_data, 'curve_identifier', error_on_failure = FALSE)
# Reorder by `Ci` and calculate average values of leaf temperature and Qin
licor_data <- organize_response_curve_data(
licor_data,
'curve_identifier',
c(),
'Ci',
columns_to_average = c('TleafCnd', 'Qin')
)What If I Have Log Files From a Licor LI-6400, PP Systems CIRAS, or Other Gas Exchange Measurement System?
At the moment, PhotoGEA can read plaintext and Excel
files created by Licor LI-6800 gas exchange measurement systems, but
does not have functions for reading log files from other
instruments.
Fortunately, there are other R packages that do, such as GasanalyzeR. Using a
function from another package is a little complicated, because generally
they do not use the same data structures or variable names as
PhotoGEA. So, after reading a file with another package,
there is typically some extra “conversion” that needs to take place.
The following example shows how to create a wrapper for the
read_6400_xls function from the gasanalyzer R
package. It also provides a simple function that “detects” the file
type, so a mix of Licor LI-6400 and LI-6800 log files could be used. The
code in the example would replace the lines that use
read_gasex_file to create licor_exdf_list in
the user guides.
As of the time this article was written (April 4, 2025), the
read_6400_xls function is not available in the CRAN version
of the gasanalyzer package. Instead, the package must be
installed directly from its GitLab repository to get the latest version.
This can be done by calling the following command from R:
remotes::install_gitlab('plantphys/gasanalyzer')This example only processes the columns that are absolutely essential for fitting A-Ci curves. Depending on your goals, you may need to modify this code to include other columns.
# Helper function for reading 6400 Excel files. This is a wrapper for the
# read_6400_xls function from the gasanalyzer package that converts the output
# into a format usable by PhotoGEA.
#
# Debug option: If DEBUG_PRINT is set to TRUE below, the raw column names will
# be printed immediately after the file is loaded. This may be helpful if there
# are other columns (such as user constants) that also need to be renamed.
read_6400_xls_wrapper <- function(fpath) {
# Read the contents of the file
rawdata <- gasanalyzer::read_6400_xls(fpath)
# Optional debug printing
DEBUG_PRINT <- FALSE
if (DEBUG_PRINT) {
print(colnames(rawdata))
}
# Rename a few columns so they meet PhotoGEA's expectations; the "new"
# PhotoGEA column name is on the left, and the "original" gasanalyzer
# column name is on the right.
new_column_names <- list(
A = 'GasEx.A',
Ca = 'GasEx.Ca',
Ci = 'GasEx.Ci',
DeltaPcham = 'Meas.DeltaPcham',
gsw = 'GasEx.gsw',
oxygen = 'Const.Oxygen',
Pa = 'Meas.Pa',
Qin = 'LeafQ.Qin',
TleafCnd = 'GasEx.TleafCnd'
)
for (i in seq_along(new_column_names)) {
gasanalyzer_name <- new_column_names[[i]]
photogea_name <- names(new_column_names)[i]
colnames(rawdata)[colnames(rawdata) == gasanalyzer_name] <- photogea_name
}
# Drop units and convert to a regular data frame
rawdata_df <- as.data.frame(units::drop_units(rawdata))
# Convert to an exdf object
exdf_obj <- exdf(rawdata_df)
# Supply units for a few columns so they meet PhotoGEA's expectations, and
# return
document_variables(
exdf_obj,
c('read_6400_xls_wrapper', 'A', 'micromol m^(-2) s^(-1)'),
c('read_6400_xls_wrapper', 'Ca', 'micromol mol^(-1)'),
c('read_6400_xls_wrapper', 'Ci', 'micromol mol^(-1)'),
c('read_6400_xls_wrapper', 'DeltaPcham', 'kPa'),
c('read_6400_xls_wrapper', 'gsw', 'mol m^(-2) s^(-1)'),
c('read_6400_xls_wrapper', 'oxygen', 'percent'),
c('read_6400_xls_wrapper', 'Pa', 'kPa'),
c('read_6400_xls_wrapper', 'Qin', 'micromol m^(-2) s^(-1)'),
c('read_6400_xls_wrapper', 'TleafCnd', 'degrees C')
)
}
# Helper function that tries to read a file using PhotoGEA::read_gasex_file, but
# then uses `read_6400_xls_wrapper` if there is an error.
read_gasex_file_plus_6400 <- function(fpath) {
cat(paste0(
'\nAttempting to read `', fpath, '` using PhotoGEA::read_gasex_file\n'
))
tryCatch(
read_gasex_file(fpath),
error = function(e) {
cat(paste0(
'\nAn error occurred. Attempting to read `', fpath,
'` using gasanalyzer::read_6400_xls\n'
))
read_6400_xls_wrapper(fpath)
}
)
}
# Load each file, storing the result in a list
licor_exdf_list <- lapply(file_paths, function(fpath) {
read_gasex_file_plus_6400(fpath)
})What If My Files Don’t Have User Constants or Other Metadata?
Please see the Guide to Licor LI-6800 User Constants.