Introduction to workloopR

Welcome to workloopR

In this vignette, we’ll provide an overview of core functions in workloopR. Other vignettes within the package give more details with respect to specific use-cases. Examples with code can also be found within each function’s Help doc.

workloopR (pronounced “work looper”) provides functions for the import, transformation, and analysis of muscle physiology experiments. As you may have guessed, the initial motivation was to provide functions to analyze work loop experiments in R, but we have expanded this goal to cover additional types of experiments that are often involved in work loop procedures. There are three currently supported experiment types: work loop, simple twitch, and tetanus.

Analytical pipelines

To cut to the chase, workloopR offers the ability to import, transform, and then analyze a data file. For example, with a work loop file:

library(workloopR)

## import the workloop.ddf file included in workloopR
wl_dat <-read_ddf(system.file("extdata", "workloop.ddf", 
                              package = 'workloopR'),
                  phase_from_peak = TRUE)

## select cycles 3 through 5 using a peak-to-peak definition
wl_selected <- select_cycles(wl_dat, cycle_def = "p2p", keep_cycles = 3:5)

## run the analysis function and get the full object
wl_analyzed <- analyze_workloop(wl_selected, GR = 2)
## for brevity, the print() method for this object produces a simple output
wl_analyzed
#> File ID: workloop.ddf
#> Cycles: 3 cycles kept out of 6
#> Mean Work: 0.00308 J
#> Mean Power: 0.08474 W
## but see the structure for the full output, e.g.
#str(wl_analyzed)

## or run the analysis but get the simplified version
wl_analyzed_simple <- analyze_workloop(wl_selected, simplify = TRUE, GR = 2)
wl_analyzed_simple
#>   Cycle        Work  Net_Power
#> a     A 0.002785397 0.07639783
#> b     B 0.003147250 0.08661014
#> c     C 0.003305744 0.09122522

Batch processing of files within a directory (e.g. successive trials of an experiment) is also readily achieved:

## batch read and analyze files included with workloopR
analyzed_wls <- read_analyze_wl_dir(system.file("extdata/wl_duration_trials",
                                               package = 'workloopR'),
                                   cycle_def = "p2p",
                                   keep_cycles = 2:4,
                                   phase_from_peak = TRUE
                                   )

## now summarize
summarized_wls <- summarize_wl_trials(analyzed_wls)
summarized_wls
#>         File_ID Cycle_Frequency Amplitude  Phase Stimulus_Pulses
#> 1 01_4pulse.ddf              28      3.15 -24.36               4
#> 2 02_2pulse.ddf              28      3.15 -24.64               2
#> 3 03_6pulse.ddf              28      3.15 -24.92               6
#> 4 04_4pulse.ddf              28      3.15 -24.64               4
#>   Stimulus_Frequency      mtime     Mean_Work   Mean_Power
#> 1                300 1728971182  0.0028362363  0.078967198
#> 2                300 1728971182  0.0009686570  0.026247519
#> 3                300 1728971182 -0.0001310863 -0.004017894
#> 4                300 1728971182  0.0024082708  0.066959552

Sections below will give more specific overviews.

Data import

Data that are stored in .ddf format (e.g. generated by Aurora Scientific’s Dynamic Muscle Control and Analysis Software) are easily imported via the function read_ddf(). Two additional all-in-one functions (read_analyze_wl() and read_analyze_wl_dir()) also import data and subsequently transform and analyze them. More on those functions later!

Importing via these functions generates objects of class muscle_stim, which are formatted to work nicely with workloopR’s core functions and help with error checking procedures throughout the package. muscle_stim objects are organized to store time-series data for Time, Position, Force, and Stimulation in a data.frame and also store core metadata and experimental parameters as Attributes.

We’ll provide a quick example using data that are included within the package.

library(workloopR)

## import the workloop.ddf file included in workloopR
wl_dat <-read_ddf(system.file("extdata", "workloop.ddf", 
                              package = 'workloopR'),
                  phase_from_peak = TRUE)

## muscle_stim objects have their own print() and summary() S3 methods
## for example:
summary(wl_dat) # some handy info about the imported file
#> # Workloop Data: 3 channels recorded over 0.3244s
#> 
#> File ID: workloop.ddf
#> Mod Time (mtime): 2024-10-15 05:46:22.352911
#> Sample Frequency: 10000Hz
#> 
#> data.frame Columns: 
#>   Position (mm)
#>   Force (mN)
#>   Stim (TTL)
#> 
#> Stimulus Offset: 0.012s
#> Stimulus Frequency: 300Hz
#> Stimulus Width: 0.2ms
#> Stimulus Pulses: 4
#> Gear Ratio: 1
#> 
#> Cycle Frequency: 28Hz
#> Total Cycles (L0-to-L0): 6
#> Amplitude: 3.15mm

## see the first few rows of data stored within
head(wl_dat)
#> # Workloop Data: 3 channels recorded over 6e-04s
#> File ID: workloop.ddf
#> 
#>    Time Position    Force Stim
#> 1 1e-04 0.503939 304.8715    0
#> 2 2e-04 0.506197 305.5165    0
#> 3 3e-04 0.505552 304.8715    0
#> 4 4e-04 0.506197 304.3875    0
#> 5 5e-04 0.505229 305.1940    0
#> 6 6e-04 0.506842 305.0330    0

Attributes

Again, important object metadata and experimental parameters are stored as attributes. We make extensive use of attributes throughout the package and most functions will update at least one attribute after completion. So please see this feature of your muscle_stim objects for important info!

You can use attributes on an object itself (e.g. attributes(wl_dat)), but we’ll avoid doing so because the printout can be pretty lengthy.

Instead, let’s just look at a couple interesting ones.

## names(attributes(x) gives a list of all the attributes' names
names(attributes(wl_dat))
#>  [1] "names"              "class"              "row.names"         
#>  [4] "stimulus_frequency" "cycle_frequency"    "total_cycles"      
#>  [7] "cycle_def"          "amplitude"          "phase"             
#> [10] "position_inverted"  "units"              "sample_frequency"  
#> [13] "header"             "units_table"        "protocol_table"    
#> [16] "stim_table"         "stimulus_pulses"    "stimulus_offset"   
#> [19] "stimulus_width"     "gear_ratio"         "file_id"           
#> [22] "mtime"

## take a look at the stimulation protocol
attr(wl_dat, "protocol_table")
#>   Wait.s    Then.action    On.port                 Units Parameters
#> 1   0.00 Stimulus-Train Stimulator .012, 300, 0.2, 4, 28         NA
#> 2   0.01      Sine Wave Length Out             28,3.15,6         NA
#> 3   0.00 Stimulus-Train Stimulator             0,0,0,0,0         NA
#> 4   0.10           Stop                                          NA

## at what frequency were cyclic changes to Position performed?
attr(wl_dat, "cycle_frequency")
#> [1] 28

## at what frequency were data recorded?
attr(wl_dat, "sample_frequency")
#> [1] 10000

Data from files that are not of .ddf format

Data that are read from other file formats can be constructed into muscle_stim objects via as_muscle_stim(). Should you need to do this, please refer to our vignette “Importing data from non .ddf sources” for an overview.

Transformations and corrections to data

Prior to analyses, data can be transformed or corrected. Transformational functions include gear ratio correction (fix_GR()) and position inversion (invert_position()). The core idea behind these two functions is to correct issues related to data acquisition.

For example, to apply a gear ratio correction of 2:

## this multiples Force by 2
## and multiplies Position by (1/2)
wl_fixed  <- fix_GR(wl_dat, GR = 2)

# quick check:
max(wl_fixed$Force)/max(wl_dat$Force)       #5592.578 / 2796.289 = 2
#> [1] 2
max(wl_fixed$Position)/max(wl_dat$Position) #1.832262 / 3.664524 = 0.5
#> [1] 0.5

A particularly important transformation - select_cycles()

Another ‘transformational’ function is select_cycles(), which subsets cycles within a work loop experiment. This is a necessary step prior to analyses of work loop data: data are labeled by cycle for use with analyze_workloop().

Data analytical functions

Core analytical functions include analyze_workloop() for work loop files and isometric_timing() for twitches. analyze_workloop() computes instantaneous velocity, net work, instantaneous power, and net power for work loop experiments on a per-cycle basis. isometric_timing() provides summarization of twitch kinetics.

To see more details about these functions, please refer to “Analyzing work loop experiments in workloopR” for work loop analyses and “Working with twitch files in workloopR” for twitches.

Some functions are readily available for batch processing of files. The read_analyze_wl_dir() function allows for the batch import, cycle selection, gear ratio correction, and ultimately work & power computation for all work loop experiment files within a specified directory. The get_wl_metadata() and summarize_wl_trials() functions organize scanned files by recency (according to their time of last modification: ‘mtime’) and then report work and power output in the order that trials were run.

This ultimately allows for the time_correct() function to correct for degradation of the muscle (according to power & work) over time, assuming that the first and final trials are identical in experimental parameters. If these parameters are not identical, we advise against using this function.

Thanks for reading!

Please feel free to contact either Vikram or Shree with suggestions or code development requests. We are especially interested in expanding our data import functions to accommodate file types other than .ddf in future versions of workloopR.