Flow Reporting and Export

Getting numbers, plots, and files out of a flow analysis, and recording the context that makes them mean something.

Gate Statistics

Every gate carries its own statistics. Open a gate's ⋮ menu and choose Show Stats.

The dialog box is titled with the gate, its event count, and its note if it has one. Below that, a row per parameter with the statistics as columns. Seven are shown by default (mean, median, geometric mean, standard deviation, %CV, minimum, and maximum) and the Columns section of the ⋮ menu turns on the rest, including quartiles and percentiles. Your choice is remembered.

Two things are highlighted for you. A maximum sitting at the top of the detector's range is flagged as saturated, and an unusually high %CV is flagged too, since both usually mean the number below them shouldn't be trusted.

Time and event-number parameters are left out, because their statistics say nothing about your cells.

That workbook gives each population its own section with its event count and percentages, its note where one exists, and its full parameter-by-statistic table. A quadrant gate contributes five sections: itself, and one for each of its four corners. A corner is named for the two conditions that define it, so a CD4 against CD8 quadrant gives you CD4/CD8 CD4+ CD8+, CD4/CD8 CD4- CD8+, and so on. You can read which population a row is without going back to the plot. Cells are shaded to carry the same warnings you see on screen: saturation and high variation in red, empty gates and gates with very few events in their own shades, and gate notes in blue, with a legend at the bottom.

Every Population at Once

Statistics Table, in the Analysis section of the sidebar, is the whole hierarchy on one screen: a row per population including All Events, and a column per statistic. It opens on counts, percentage of parent, and percentage of total, plus the mean and median of every fluorescence channel. Scatter channels are left off, because their averages rarely say anything.

Columns adds and removes statistics and channels. Click a column heading to sort by it, which is the quick way to find the tube's largest population or its dimmest marker. Distribution is worth adding: it draws each population's histogram for that channel right in the cell, so you see the shape behind the number rather than just its average. A twelve-colour panel opens with two dozen columns, so the panel also carries Clear all, which strips the table back to population names, and Restore defaults, which puts the opening set back.

The ⋯ menu holds everything else. Copy table puts the whole thing on the clipboard for pasting into a spreadsheet, Export CSV writes it to a file, and Send to Data Table files it in the lab record. All three take every population; to report on a few, take the whole table out and cut it down there.

The button waits until you have drawn a gate, and says so if you hover it.

Asking Whether Two Populations Differ

Compare Populations, below the table, takes two populations and one channel and runs a Kolmogorov-Smirnov test on their distributions. You get the D statistic and a p-value, and a note when the p-value falls below 0.05.

The test compares the shapes of two distributions rather than their averages, so it catches a population that has split into two peaks where a difference in means would not. It needs two gates before it will open.

Seeing Which Markers Move Together

Correlation Heatmap, below the comparison, plots every fluorescence channel against every other one and colours the grid by how tightly the pair tracks. Red is positive, blue is negative, and a pale grey is no relationship. A scale under the grid maps the colours back to numbers.

Hovering a square names both channels and gives you the coefficient in the bar above the grid. Tapping does the same on a phone, and once you have clicked into the grid the arrow keys move one square at a time.

Population narrows the calculation to the cells inside one gate, so you can ask whether two markers travel together in your CD4 cells specifically rather than across the whole tube. Pearson measures a straight line relationship and Spearman measures rank order, which holds up better on a skewed channel.

Parameters opens the channel list, and the button carries the count of how many are switched on. Two have to be on before the grid draws. A panel past about a dozen channels gets a Filter channels box at the top of the list. Scatter and time channels are left out entirely, since correlating cell size with acquisition order says nothing about your panel. Export CSV writes the matrix to a file.

A channel that never varies has no correlation to report. Its squares are hatched and read n/a, so a dead detector cannot be mistaken for a clean negative result.

It reads whatever the plots are showing, so an analysis with compensation applied correlates compensated values. A loaded file is all it needs, no gates.

Coefficients come from up to 10,000 events, taken at an even spacing across the whole file rather than from the first 10,000, which is enough for a correlation and keeps the grid quick to redraw. When a file runs past that cap, a line under the controls says how many events went in and how widely they were spaced, and the exported CSV carries the same line at the top. A file under the cap is read whole, so no line appears.

Comparing Across Files

Two tables handle the cross-file case, both reached from the ⚙ on the Files section.

Batch Statistics comes out of Compare, with a row per file and three columns per population: count, percentage of parent, and percentage of total, plus a mean and standard deviation footer. A quadrant contributes five: itself, and one for each of its four corners, each named for the two conditions that define it, such as CD4/CD8 CD4+ CD8+. Very low and very high percentages are tinted so an obviously failed file stands out, and any file where a gate couldn't be placed is called out rather than left looking like a zero.

Aggregate Statistics comes out of Statistics, and adds condition banding, outlier flagging, and significance testing. It's described in Flow files and templates.

Sending Statistics to a Data Table

Every one of those screens offers Send to Data Table, which writes the numbers into a real table in Data Tables, named after the analysis. From there you can chart it, run a test on it, join it to something else, or export it as Excel.

The table is a snapshot, taken when you send it. Re-gating afterwards doesn't rewrite it, which is deliberate: a table you've built a figure on shouldn't change underneath you because someone nudged a gate. Send it again to get the current numbers.

Sending to a table and adding to a figure both create something, so they're not offered to project members with view-only access. Exporting a CSV or a spreadsheet is available to everyone.

Charts and Figures

You don't build charts inside the flow workspace. Statistics are kept as a table behind the analysis, and you chart that table in Figures, which is where everything else in Conspecta gets composed.

That table covers every file loaded in the analysis, not just the one you're looking at, so a chart from a multi-file analysis compares your samples without you sending anything anywhere. The file you have open carries its full set of marker statistics, and the others carry counts and percentages, which is what the background comparison measures. A quadrant's four corners each get their own row, so a double-positive frequency charts across samples like any other population.

Adding a chart from a flow analysis offers ready-made views:

• Frequency (% of parent) or Event count by population
• Population composition as a stacked proportion
• Median intensity heatmap across markers
• A specific marker's median by population
• Any dot plot from your analysis, drawn live from the FCS data

The Batch Statistics screen also has Add to figure directly, which is the short path from a comparison to a panel.

Add to figure on a plot's ⚙ menu does the same for a single dot plot: pick an existing figure or name a new one, and the plot lands as a panel. The panel stays live. It redraws from the FCS data, so re-gating the plot updates the figure. Project members with view-only access don't see the action, since adding a panel creates something.

Exporting Plots and the Tree

Export plot on a plot's ⚙ menu offers PNG or SVG, either the current plot or every plot at once, at one, two, or three times screen resolution.

SVG carries the extras: a title, axis labels, font and background choices, and whether gates are drawn on it. PNG is a straight capture of what's on screen, so set the plot up the way you want it before exporting.

Export Tree in the Analysis section opens the figure workspace, with the figure itself on screen and the choices beside it. Export in the left rail asks what to capture: the whole layout, the selected plot, or a contact sheet, every plot tiled into one image, sized for a single slide rather than spread across the gating tree's branches. Each choice says what it would produce, and the picture redraws as you change it, so you see a contact sheet before you commit to one. Then the format, whether to hide the boxes, the file name, and a white or transparent background.

Size is asked the way a journal asks for it: a Width of Single column (89 mm), Double column (183 mm), Slide (254 mm), or Custom in millimetres, plus a Resolution of 150, 300, or 600 DPI. The panel shows the pixel dimensions those choices produce, so you know the file meets a submission requirement before you write it. An SVG prints at the chosen width at any resolution. Export PNG (or Export SVG) writes the file and confirms it by name, since a browser download can be easy to miss. Copy to clipboard puts a PNG straight on your clipboard for pasting into a slide, and the button says Copied when it has.

Annotate, the panel below Export, draws on the figure itself. An arrow, a circle, or a label you add is saved with the analysis and comes back the next time you open it, and in an SVG export it arrives as a shape you can still move and recolour rather than pixels burned into a picture. Changing what you capture never discards them, so switching to a single plot and back leaves your marks where you put them.

SVG is available for all three, the whole gating tree included. It is built rather than photographed. Each plot's events stay a raster layer, because there are far too many to draw one by one, and everything you would want to restyle stays vector on top of them: gate outlines, gate labels, the connector arrows, and each connector's gate name and percentage. Open it in Illustrator or Inkscape and you can retype a gate name, recolour an outline, or move a label without touching the data underneath.

The file name is filled in for you from the analysis, what you captured, and today's date (T-cell-panel_gating-tree_2026-08-04), so a second export doesn't land as a copy of the first and a folder of them sorts into order. Type over it whenever you want something else.

Transparent backgrounds suit a slide with a colour behind it. Word and PowerPoint render transparency as black, so pick white for those.

Hide boxes is worth knowing what it does. On screen each plot sits in a card with a border and a file name above it, which reads well in the workspace and badly in a manuscript, where a figure panel wants the plot on its own. Tick it and the export drops that chrome. The picture beside the choices redraws either way, so you can see both before you commit.

Exporting Data

Export FCS on a gate's menu writes that population out as a real FCS file, with the gate's name recorded in the file's keywords. That's how you hand a sorted population to a collaborator or a tool that isn't Conspecta.

The whole workspace's raw files and gating export from Settings → Export under Flow Cytometry, which produces a ZIP of the FCS files and the gating hierarchies.

Recording Context

Experiment Metadata records what the numbers were measured on: experiment date, instrument, operator, condition, timepoint, patient identifier, and the staining panel. Click the analysis name at the top of the workspace to open Analysis Details, where it sits below the notes. Fill it in once and it travels with the analysis.

Those seven are offered because most panels want them, and none of them is required. Type a name and a value in the row at the bottom to add a field of your own, and hover a field you added to remove it. You get 50 fields in all, with names up to 64 characters and values up to 500. Keywords save with the rest of the dialog, so Save commits them and Cancel puts them back.

Gate notes are the other half of that. Add one from any gate's ⋮ menu to record why the gate sits where it does, such as "set at FMO + 2 SD" or "based on isotype control". Notes appear on the gate's badge, in its statistics, in the gating path above the plots, and in spreadsheet exports, which means the reasoning survives the person who did the gating.

Data-Quality Warnings

Conspecta checks each analysis as you work and surfaces two warnings above the plots: a spillover matrix present but not applied, and a file with fewer than 1,000 events in total.

Finer-grained checks appear where they're relevant rather than as banners. Saturated parameters and high variation are flagged in the statistics cells themselves, and empty or nearly empty gates are flagged on the gate. All of them are informational and none of them block you.