IsoFishR 2 (development version)

Data reduction and analysis of ⁸⁷Sr/⁸⁶Sr ratios measured by laser-ablation MC-ICP-MS, for example along otolith transects. IsoFishR turns instrument runfiles into blank-corrected, Rb-interference- and mass-bias-corrected profiles. You then trim them, mark regions, detect changepoints and export the results, all in an app that runs in your browser.

Looking for the original IsoFishR (v1)? It is still available as the v1.1 release, as described in Willmes et al. 2018, PLOS ONE.

Installation

You need R (4.1 or newer). RStudio is optional. In R, run:

install.packages("IsoFishR", repos = c("https://maltewillmes.r-universe.dev", "https://cloud.r-project.org"))

Run the same line again to update. No extra build tools are needed on Windows, macOS or Linux.

Start the app

IsoFishR::run_app()

The app opens in your web browser. To stop it, click Quit IsoFishR (top right). If you close or reload the tab instead, the browser asks first; a closed tab stops the app after a few seconds.

The six steps

The app guides you through a project from left to right. Quit IsoFishR and the Session log are always at the top right. Everything is saved as you work, and a project reopens on the step you last worked on.

1. Project. Create a new project (a folder you choose, e.g. in Dropbox or on a shared drive) or open an existing one; recent projects are listed. Pick the instrument preset when creating a project.
2. Settings. Check the runfile layout first.
   • Instrument preset: Nu_UCDavis (Nu Plasma) is built in. For another instrument, set the layout, test it with Check the layout with a runfile, and save it as a preset to reuse it.
   • Reduction defaults: run speed, integration, blank time, 88Sr limits, outlier rejection and smoothing: none, moving average, spline, or moving average + spline (default; both lines are drawn, and the spline is the smoothed value in summaries). The spline sets its own smoothness from the data: noisy profiles get a smoother curve, and it can still follow sharp transitions.
3. Runfiles. Add runfiles as individual files, or choose a folder: all files with the preset's extension (.run) in it and its subfolders are found.
   • More batches: you can add more at any time.
   • Same name as an existing run: choose whether to keep the existing run or replace it with the new file. Replaced runs start afresh: their edits, regions and notes are removed (the old file is kept in raw/removed).
   • Damaged files: a file whose header was saved wrongly (lines missing or added) is still read: IsoFishR finds where the data starts and notes it. Files without readable data are skipped, and a summary lists which files were added, skipped or failed, and why.
   • Runs table: lists all runs in acquisition order (read from the runfile timestamps), and lets you remove runs.
4. Calibration. Correct the samples to a reference material measured in the same session, in the style of Iolite.
   • Default reference materials: runs with SBO or GSFR in their name are suggested as reference material runs (accepted value 0.70918 for both, which you can change). The first one marked becomes the primary reference material.
   • Other reference materials: add them with their accepted value and ± 2σ, then mark their runs under Assign runs. Runs can also be marked as baselines (gas blanks).
   • Primary vs secondary: the primary reference material corrects the samples. By default this is the mean of its runs within each analytical session. A new session starts after a gap of more than 6 hours, and runs can be moved between sessions. The correction is applied to every point at the time it was measured. For instrument drift, choose a linear regression through time, joining neighbouring standards, a smooth curve, or the most recent standard. All other reference materials are secondary: they are corrected like samples and reported per session as accuracy checks.
   • The plot: each reference material's run means before and after correction, the regression line, the accepted values and the secondary session averages.
   • The table: uncorrected and corrected mean and SD per reference material and session, with the deviation from the accepted value.
5. Workspace. Large plots of ⁸⁷Sr/⁸⁶Sr, total Sr, ⁸³Kr and ⁸⁵Rb/⁸⁸Sr for each sample. A badge shows whether the values are corrected to the reference material or uncorrected. When the calibration changes, the plots update.
   • Selecting a range: drag across the plot. The footer shows the range's mean ± 2 SE. Zoom with the mouse wheel; double-click to reset.
   • Custom lines: enter 87Sr/86Sr values (e.g. 0.7092, 0.7105) above the plot to draw labelled lines on every sample of the project.
   • Edit: trim, add the selection as a named region, reverse the profile, or detect changepoints.
   • Reduction: change the run speed, 88Sr limits, outlier rejection or moving average window for one sample only.
   • Notes: status, flag and comment. "Mark reviewed and go to next" steps through the samples.
6. Export. One button writes all results (excluded samples left out) to a new dated folder inside the project; earlier exports are kept. From R, export_project() can write a chosen subset.

Each sample keeps its own SE. The uncertainty of the correction, from the primary measurements and the accepted value, is reported separately and in a combined SE. Without a primary reference material, all values stay uncorrected.

To try IsoFishR without your own data, add the example runfiles that come with it (five Chinook salmon otoliths and one white seabass otolith). Find their folder with system.file("extdata", "runfiles", package = "IsoFishR").

What is in a project folder

MyProject/
  project.yml   project settings, including the import preset
  raw/          copies of the original runfiles (never changed)
  samples.csv   per run: type, acquisition time, session, status, flag, comment, trims, setting overrides
  regions.csv   regions per sample
  reference_materials.csv   reference materials with accepted values
  log.csv       every change: when, who, what, old and new value
  cache/        reduced data (safe to delete; rebuilt automatically)
  exports/      one dated folder per export
  app_state.yml the step the project was last worked on

Projects can be shared by sharing the folder. Only one person can edit a project at a time; others can open it read-only.

Exports

Each export creates exports/<date_time>/ containing:

File	Contents
profiles.csv	Every point of every profile in one table: distance, time, 87Sr/86Sr, moving average and spline (with SE), region, changepoint segment, and signal intensities
samples.csv	One row per sample: type, session, status, comment, edits, profile mean, SD and SE (and, when corrected, the uncorrected mean, correction factor and combined SE)
regions.csv	One row per sample and region: mean, SD, n, SE, distance range
changepoints.csv	One row per changepoint segment
runs.csv	All runs in acquisition order with type and session
reference_materials.csv	Every reference material run: measured and corrected mean, deviation from the accepted value (ppm), whether it is within uncertainty
reference_material_summary.csv	Per reference material and session: runs, mean, 2 SD, deviation
settings.csv	Every setting each sample was reduced with, including the calibration
export_report.md	Always written: a readable record of the export (Markdown). Date, user and IsoFishR version; runs and sessions; the calibration method and baseline mode; the primary and each secondary reference material with accepted value, number of runs and mean and SD before and after correction (overall and per session); all project settings; and per-sample setting overrides

Using IsoFishR from R code

All of the app's functions are available in R scripts, for example:

library(IsoFishR)
p <- create_example_project()
p <- set_regions(p, "PC_2016_01_A2_4R",
                 data.frame(region = "core", start_s = 200, end_s = 260))
summarise_project_regions(p)
export_project(p)
close_project(p)

Citation

Willmes M, Ransom KM, Lewis LS, Denney CT, Glessner JJG, Hobbs JA (2018). IsoFishR: An application for reproducible data reduction and analysis of strontium isotope ratios (⁸⁷Sr/⁸⁶Sr) obtained via laser-ablation MC-ICP-MS. PLOS ONE 13(9): e0204519. https://doi.org/10.1371/journal.pone.0204519

License

MIT. Questions and bug reports: GitHub issues.