One row = one data file

Each row in an SDRF file represents one data file (e.g. a .raw or .mzML file) and the biological sample it came from. If the same sample was run twice, it gets two rows. If a TMT multiplex packs 10 samples into one file, that file gets 10 rows — one per channel.

This design makes the relationship between samples and data explicit, so analysis tools can automatically match files to conditions without manual configuration.

Three types of columns

Every SDRF column belongs to one of three categories. Understanding these makes filling in a template straightforward.

Use controlled vocabulary

SDRF values should come from established ontologies whenever possible. This ensures that "breast cancer" in your file means the same thing as "breast cancer" in someone else's file, enabling cross-study analysis.

• Organisms — use NCBI Taxonomy names: Homo sapiens, Mus musculus
• Diseases — use MONDO or EFO terms: breast cancer, normal
• Instruments — use PSI-MS terms: Orbitrap Exploris 480, Q Exactive HF
• Modifications — use UNIMOD names: Oxidation, Carbamidomethyl
• Tissues — use UBERON or BTO terms: liver, blood plasma

The SDRF Terms page lists all allowed values and their source ontologies. The SDRF Editor provides autocomplete for ontology terms.

Start from a template

You don't need to build an SDRF from scratch. Templates define which columns are required and recommended for your experiment type. They are organized in layers:

• Organism templates (human, vertebrates, plants) add species-specific metadata like age, sex, ancestry, or growth conditions.
• Experiment templates (DIA, single-cell, crosslinking) add technique-specific columns like isolation windows or crosslinker details.
• Templates are composable — combine an organism template with an experiment template to get all the right columns.

Browse available templates on the home page or use the Template Builder to interactively pick the right combination.

Factor values tell tools what you're comparing

Factor value columns are what make an SDRF file useful for analysis. They tell tools like quantms which sample properties differ between your experimental groups.

If you're comparing healthy vs. disease samples, add factor value[disease] with the same values as characteristics[disease]. If you're comparing treatment vs. control, add factor value[compound]. Multiple factor values are allowed when comparing more than one variable.

Common patterns

Validate before submitting

The sdrf-pipelines validator checks your file against the specification: required columns, valid ontology terms, consistent labels across rows, and more.

Install and run:

pip install sdrf-pipelines
parse_sdrf validate-sdrf --sdrf_file your-file.sdrf.tsv

Fix any errors reported, then validate again. A passing validation means your SDRF is ready for submission to PRIDE or other ProteomeXchange repositories.