Prepare an analysis-ready dataset

Before you begin

Keep an unchanged copy of the instrument export. Prepare a working copy for cleaning and record which sheet, rows, and columns you changed. Use stable sample identifiers so that a result can be traced to its source measurement.

Choose the workflow from the data you have: CSV or Excel measurements for assays, individual event files for flow gating, or exported population tables for flow summary analysis. A population summary cannot reconstruct the original events.

Describe the experimental design

Include the treatment or condition, sample identifier, measurement, and units. Add donor, plate, time point, or batch identifiers when those distinguish observations. Keep control samples explicitly labeled rather than encoding their role only in a filename.

Tell the agent what one independent sample represents and which measurements belong together. Repeated wells, repeated measurements of a donor, and different donors should have distinguishable identifiers. Specify any pairing in the request; adjacent rows alone do not establish a pair.

Example: a traceable measurement table

This illustrative layout keeps the biological sample and technical replicate separate. Adapt the column names to your experiment; these example measurements are for learning the workflow.

sample_id,condition,technical_replicate,response,unit
S01,control,1,102,RFU
S01,control,2,105,RFU
S02,treated,1,74,RFU
S02,treated,2,78,RFU

State the analysis request

A useful request names the input, mapping, calculation, and requested output. For example: “Use the Results sheet. response is in RFU. sample_id identifies a biological sample; technical_replicate identifies repeat wells. Summarize repeat wells within each sample and show a per-sample table before comparing conditions.”

If you have concentrations, state whether they are original or log-transformed, and identify their units. For dilution series, state the dilution convention and any factor already applied. Do not rely on a number such as 10 to communicate whether it means a concentration or a dilution.

Handle missing and excluded measurements

Keep missing, not-detected, and out-of-range observations distinguishable in your working record. Do not replace them with zero just to make the table numeric. Tell the agent what those labels mean and inspect which observations enter the analysis.

Document each exclusion with a sample or well identifier and a reason. If you ask for an alternative analysis with exclusions, preserve the original result first so the effect can be reviewed.

Verify the import before starting

1. Compare the displayed file and sheet names with the source export.
2. Check column mappings, units, sample labels, and the number of observations.
3. For a plate, verify row and column orientation and assign blanks, controls, standards, and unknowns.
4. Check that the prompt describes normalization, grouping, pairing, and exclusions.
5. Resolve any question about the data mapping before interpreting a result.

Completion criteria

You should be able to identify the source of every analyzed measurement, explain the role of each column, and describe any preprocessing. Continue with Uploading Data and the relevant assay guide once those checks pass.