CodeBlock example — run an R script (AccuCor)

accucor.R runs AccuCor natural-isotope correction on an LC-MS peak table. CodeBlock is how you bring an existing Python, R, or Julia script into a workflow without rewriting it as a Python block — ideal when the science already lives in a language-specific package (here, the R accucor package).

How a CodeBlock works

Unlike the other base classes, you do not subclass anything. A CodeBlock is a script plus a small port configuration that says what files go in and out. SciStudio:

1. Creates an exchange folder with one sub-folder per port.
2. Writes each connected input into inputs/<port>/ in the file format you declared.
3. Runs your script, telling it where the folders are via environment variables.
4. Reads each file your script left in outputs/<port>/ back into a typed DataObject.

The script learns the folder locations from environment variables SciStudio sets:

Variable	Points to
SCISTUDIO_INPUTS_DIR	the inputs/ folder (one sub-folder per input port)
SCISTUDIO_OUTPUTS_DIR	the outputs/ folder (write one sub-folder per output port)
SCISTUDIO_EXCHANGE_DIR	the exchange root
SCISTUDIO_SCRIPT_PATH	this script's own path

The port configuration

This block declares one input and one output port. Conceptually:

Port	Direction	Data type	Extension	Folder
peaks	input	DataFrame	.csv	inputs/peaks/
corrected	output	DataFrame	.csv	outputs/corrected/

So the script reads its peak table from inputs/peaks/*.csv and writes the corrected MID table to outputs/corrected/mid_table.csv — exactly what accucor.R does. SciStudio then turns that CSV back into a DataFrame on the corrected output port.

You set this configuration when you add the CodeBlock and point it at the script (script_path), choosing the language by the script's extension (.R, .Rmd, .qmd for R; .py for Python; and so on). The exact config model is CodeBlockConfig / PortFileConfig in scistudio.blocks.code — see the API reference.

What AccuCor does

LC-MS isotope-tracing experiments measure metabolite peaks, but every peak carries a background of naturally heavy isotopes (≈1.1% of carbon is ¹³C even with no tracer). AccuCor subtracts that natural abundance, leaving the tracer-derived mass-isotopomer distribution (MID) — the real signal that downstream flux analysis needs.

Why use a CodeBlock instead of a Python block

• The logic already exists in R/Julia and you do not want to port it.
• A domain package ships a validated script you want to reuse as-is.
• You are prototyping in a notebook-style language.

If the logic is small and Python-native, a ProcessBlock is simpler — keep CodeBlock for genuinely cross-language work.