Targets one base apart
Even an exact read can fit both targets under radius-one matching. A best-distance policy would treat an exact match differently.
Free research tool · No installation · No upload
Can a one-base correction confuse your targets?
Inspect duplicate sequences, exact-read ambiguity, and overlapping single-substitution neighbourhoods. Your library is processed in this browser, without sending sequences to a server.
This is an assignment-collision checker, not a CRISPR off-target cutting, editing-efficiency, or clinical-safety assessment.
Check your target library
What is counted?
For every supplied sequence of length L, the audit enumerates the exact sequence and all 3L single-base substitutions. An observation is ambiguous when it lies within one substitution of more than one supplied target ID. Shared observations are counted once, not once per conflicting pair.
Targets two bases apart
The exact target sequences remain distinct, but their one-substitution neighbourhoods overlap.
Targets at least three bases apart
Their radius-one Hamming neighbourhoods are disjoint. This is a statement about the supplied library and matching model, not the accuracy of an entire assay.
A 20-base target has 60 possible single-substitution observations. The number that is ambiguous is not an estimated error probability: actual sequencing errors are not assumed to be equally likely.
Limits that matter.
This audit uses all supplied target rows, supplied orientation, A/C/G/T sequences of equal length, and Hamming radius one. Duplicate sequences remain separate candidates when their IDs differ. No automatic deduplication, sequence repair, reverse-complement search, indel model, or quality weighting is applied.
A unique assignment can still be wrong because the true target is missing, contamination is present, or the error process exceeds the model. No green score or biological confidence percentage is inferred from collision-free geometry.
The browser limit is 2,000 targets of 8–32 bases. Large libraries are rejected, not silently sampled. Only the displayed table and witness list are shortened; the completed JSON report contains every target’s audit.
Move from a library check to an assay workflow.
Use the native audit for a larger library and review its configured metric and semantics. This small exhaustive implementation is independent of the native engine and is not a native-performance benchmark.
dotmatch audit --help
dotmatch audit \
--targets guides.tsv \
--k 1 \
--audit-mode auto \
--out-dir audit/