Exporting

Exporting requires a license; the 30-day trial includes it. Saving a workspace does not — it is never blocked.

Statistics as CSV

Export in the statistics panel offers two layouts, for the active file or — in the All files view — for every open file:

Choice	What you get
This table	One row per population, with the statistic columns you have on screen
All parameters	One row per population and parameter, with every statistic — for marker sweeps

Every row starts with File Name, Path, Gate Name, Type, X axis, Y axis, Count, % of Total and % of Parent. Count and both percentages are always included, even if you have hidden them on screen.

The Compensation column

The last column of every row, Compensation, records how that row’s file was compensated, because a multi-file export can mix compensated and raw files and nothing else in the file says which:

Value	Meaning
on	Compensated with the file’s own acquisition matrix
on*	Compensated with a different matrix — edited, imported, computed from controls, or applied from another file
off	Not compensated: raw values
mismatch	Compensation is on, but the matrix names channels the file does not have

These match the Comp, Comp* and Comp ⚠ badges in the file list — see Compensation. A file whose statistics cannot be computed — a mismatch file is one — is left out of the export, and the app tells you which: “Exported … of … files. No statistics for: …”.

It is last so that a script reading the earlier columns by position keeps working.

Blank means “not counted”, not zero

A population that cannot be counted is written with empty cells in the This table layout, never as 0. A real zero and “could not be computed” are different results, and an empty cell keeps them apart.

For an experiment rather than a single sample, batch analysis can also export a files × gates table.

How CSV files are encoded

Every CSV the app writes — statistics, events, batch, dose-response and the compensation matrix — is UTF-8 with a byte-order mark (BOM).

The mark is what makes Excel open the file correctly. Without it, Excel on Windows reads a CSV in the system’s legacy code page — Shift-JIS on Japanese Windows, for example — and every non-ASCII character turns into garbage: a gate name typed in Japanese, a marker name with a Greek letter, R² in a dose-response fit. With the mark, Excel reads the file as UTF-8 and the names come through intact.

Column headers are always in English, whatever language the interface is in, and numbers always use a full stop as the decimal separator. The files are data: a script or a colleague reads them the same way everywhere.

If you read the files in code, use an encoding that drops the mark:

• Python — open(path, encoding="utf-8-sig"). pandas’ read_csv already handles it.
• R — read.csv(path, fileEncoding = "UTF-8-BOM").

Otherwise the first column’s header can come through as File Name, and a lookup by name fails.

Event data as CSV

Export in the toolbar:

Choice	What you get
All Events (CSV)	Every event in the active file
A gate, by name	Only events inside that population — the gate and all its parents
Batch Export All Files	Every event of every open file, one CSV per file, into a folder you choose. Shown when more than one file is open.

Per-gate event export is the route to another tool — hand the events for a population to R or Python without re-deriving the gate there.

Each file’s events are exported with its own compensation. A compensated channel is labelled Comp-… (for example Comp-FITC-A), so a compensated column is never mistaken for a raw one.

Event exports are large

An event-level CSV of a million-event file is a big text file, and a batch export across an experiment is much bigger. Export the population you need rather than everything, when you can.

Each population is listed with the colour it carries in plots and in the gate list, so the population you are exporting is the one you meant.

Figures

A plot’s Export menu offers Export as PNG, Export as SVG and Export as PDF. Before saving, Export Settings asks for the width, height, resolution — 72 (Screen), 150 (Draft) or 300 (Print) — and background (transparent, white or black). Preferences → Export Defaults sets the defaults.

Plot exports are images, in every format

All three formats contain the plot as an image of what is drawn on screen, scaled to the size you choose. The SVG and PDF wrap that image; they are not vector drawings, so the dots and gate outlines cannot be edited separately in Illustrator or Inkscape, and a higher resolution enlarges the screen image rather than adding detail.

Two things follow. Make the plot window as large as you can before exporting, and remember that a plot draws a subsample of events — the footer shows how many — so very rare events may not be on it.

Graph Builder charts are different: their Export SVG is a true vector drawing, editable element by element. Their Export PDF and Export PNG are images.

The compensation matrix

Export in the compensation panel writes the spillover matrix as a CSV — a header row of channel names, then one row per channel. Do it, for the record.

Gates

• Export Gates in the gate list saves a Cytogence gate template (.cgt), which Import Gates reads back — the way to reuse a strategy in another analysis.
• Export Gating-ML writes the gate tree as ISAC Gating-ML 2.0, readable by FlowKit, Cytobank and FCS Express, with the compensation each gate was drawn in: the file’s own matrix, none, or — for an edited matrix — the matrix itself, written into the document.
• Import Gating-ML keeps each gate in the compensation it declares, asking before it switches the file’s compensation. A gate whose parent could not be imported is skipped and listed, never added as a top-level gate.

What makes a result reproducible

A statistics table on its own does not record how it was produced. For a figure or number you may need to defend, export:

1. The statistics — the numbers themselves, with the Compensation column.
2. The compensation matrix — the transform applied to the data.
3. The workspace — gates, hierarchy and each file’s compensation. See Workspaces.
4. The AI audit trail, if the copilot was involved — which steps were proposed and which you approved.
5. The AI-1 competence report, if AI-1 gated anything — including which populations it declined.

Points 4 and 5 are the ones people forget, and they are the ones a reviewer asking “how was this gated?” is actually asking about.