Compute¶

Single Table Math¶

Creates or modifies a column using a pandas eval expression.

Details

Examples:

Ratio = Intensity_Ch1 / Intensity_Ch2 — create a new column
Area_um2 = Area * 0.065 * 0.065 — convert pixels to physical units
Normalized = Intensity / Intensity.mean() — normalize to mean
Log_Area = @np.log10(Area) — use numpy functions with @ prefix

Uses pandas DataFrame.eval() syntax. The left side of = is the new column name.

Direction	Port	Type
Input	`in`	table
Output	`out`	table

Aggregate Table¶

Reduces a table to aggregate statistics across rows, optionally grouped by column.

Details

Without grouping, all numeric columns are reduced to a single row. With grouping, each unique group gets its own summary row.

Group	Area
Control	110
Treated	190

Parameters:

Operation — sum, mean, median, min, max, count, std, var, auc
Group By — column name(s) to group by (comma-separated, leave empty for no grouping)
Columns — restrict to specific columns (comma-separated, leave empty = all numeric)
Sort By — column to sort by before computing AUC (required for auc, e.g. time column)

The auc operation computes the area under the curve using the trapezoidal rule. Rows are first sorted by the Sort By column, which serves as the x-axis (e.g. time). Each selected numeric column is then integrated against that x-axis.

Direction	Port	Type
Input	`in`	table
Output	`out`	table

Properties: Operation

Two Table Math¶

Computes a scalar arithmetic operation between one value from each of two input tables.

Details

For each input table the node picks the first numeric column (or the column named in the matching property) and uses row 0 as the scalar value. Designed for comparing scalar outputs such as stained-area measurements.

Outputs a single-row result table: left_value | right_value | operation | result

Parameters:

Operation — left / right, left * right, left + right, or left - right
Left Column — column name in the left table (blank = first numeric)
Right Column — column name in the right table (blank = first numeric)

Direction	Port	Type
Input	`left`	table
Input	`right`	table
Output	`result`	table

Properties: Operation

Normalize Column¶

Normalizes one or more numeric columns.

Details

Methods:

Min-Max (0-1) — scales each column to [0, 1]
Z-score — subtracts mean and divides by std (standard score)
Log10 / Log2 / Ln — log transform (adds 1 before log to handle zeros)
Robust (IQR) — subtracts median, divides by IQR; robust to outliers

Parameters:

Columns — comma-separated names. Leave empty to normalize all numeric columns.
Suffix — text appended to new column names (e.g. _norm). Leave empty to overwrite in-place.

Direction	Port	Type
Input	`in`	table
Output	`out`	table

Properties: Method

Value Counts¶

Counts occurrences of each unique value in a column.

Details

Outputs a two-column table with the original column name and a count column, sorted by count descending by default.

Parameters:

Column — the column to count unique values in
Sort by count (descending) — sort results by frequency
Add percentage column — include a pct column with relative frequencies

Direction	Port	Type
Input	`in`	table
Output	`out`	table

Properties: ,

Group Normalization¶

Normalizes numerical columns relative to a specified control group mean.

Details

Each numeric column is divided by the mean of its corresponding control group, producing fold-change values. A mapping table widget lets you assign a different control group for each unique group in the data.

Parameters:

Global Control Group — default control group name used for normalization
Target Column — column containing group labels (e.g. Group)

Direction	Port	Type
Input	`in`	table
Output	`out`	table