Methods

Transparent set arithmetic.

VennScope maps each literal identifier to a membership signature and derives intersections from those observed signatures.

Input normalization

Blank entries are removed and duplicate identifiers within a set are deduplicated. Matching is case-sensitive by default; users may explicitly choose case-insensitive comparison. VennScope does not silently uppercase biological identifiers.

Declared identifier formats

Datasets can be labelled as gene symbols, UniProtKB accessions or entry names, Ensembl gene/transcript/protein IDs, NCBI Gene IDs, RefSeq accessions, HGNC IDs or arbitrary identifiers, together with organism metadata. Raw analysis still uses the literal identifier supplied by the user. Format checks are plausibility checks only and do not establish biological identity.

v0.16 adds a separate, explicit online mapping workflow using the UniProt ID Mapping service for supported routes. Some external-to-external conversions require two documented mapping steps through UniProtKB. VennScope identifies the mapping service and route, records the access date, displays one-to-many/unmapped records, preserves the original dataset and requires explicit permission before selected identifiers are sent to the external service.

Exact intersections

For sets A, B and C, an exact A ∩ B region contains identifiers present in A and B but not C. Exact regions correspond to one complete membership signature.

Inclusive intersections

Inclusive A ∩ B contains every identifier present in both A and B, including identifiers that are also present in C or additional sets.

Element comparison

The Elements workspace can inspect the full union or a selected intersection, filter identifiers by membership degree, and compare selected identifiers across every loaded dataset using the same membership signatures that drive the diagram.

Similarity

Jaccard = |A∩B| / |A∪B|. Sørensen–Dice = 2|A∩B| / (|A|+|B|). Overlap coefficient = |A∩B| / min(|A|,|B|).

Overlap statistics

Statistical overlap testing requires a background universe. For each dataset pair, v0.16 can report a one-sided hypergeometric upper-tail enrichment p-value and a two-sided Fisher exact p-value from the corresponding 2×2 membership table. Odds ratios are reported with approximate 95% log-scale confidence intervals; when a table contains a zero cell, a 0.5 Haldane–Anscombe correction is used for odds-ratio estimation. Benjamini–Hochberg q-values adjust the displayed pairwise Fisher p-values for multiple testing.

These tests evaluate overlap between identifier sets. VennScope does not apply a t-test to ordinary set lists because t-tests require continuous numerical measurements and an appropriate replicate design.

Diagram geometry

The Venn representation in v0.16 is schematic. Circle area and overlap area are not claimed to encode exact cardinality. Numerical labels are derived from the set calculations. For four or more sets, VennScope recommends an UpSet representation.

Independent verification

For consequential scientific use, reproduce critical calculations using an independent workflow before publication.

Identifier mapping

Raw set analysis compares identifiers literally after the selected normalisation rules. Optional v0.16 identifier mapping is a separate, explicit workflow. For supported routes, selected identifiers are submitted directly from the browser to the UniProt ID Mapping service; the original dataset is preserved, mapped/unmapped/one-to-many results are shown for review, and source provenance can be exported.

Mapping should not be interpreted as proof that gene-, transcript- and protein-level entities are interchangeable. Users should choose a common biological entity and identifier system appropriate to the scientific question.

Learning resources

The VennScope Learn section explains set theory, similarity coefficients, overlap statistics, multiple testing and biological identifiers without disclosing VennScope's proprietary implementation details.