Pooled CRISPR screens
Count known guides, inspect assignment QC and export a guide-by-sample matrix.
CRISPR countingDotMatch
CRISPR guide counting and barcode assignment
Count known guides from FASTQ files and export MAGeCK-compatible tables. Review unique, ambiguous, unmatched and invalid reads.
Apache-2.0Linux & macOSLocal processing
Read assignment exampleRadius-one Hamming
Observed read window
CGCATGCATGCATGCATGCA1 compatible target
uniqueguide_C
One target within one substitution. Adds one count to that target.
Synthetic example, checked against the native matcher. Example inputs and tests
Current source benchmark · One mismatch, no indels
DotMatch versus guide-counter
- Faster complete commands
- 1.6–5.5×
- Lower peak memory
- 11.3–14.4×
- Guide/sample count differences
- 0
The benchmarked source checkout's dotmatch guide-counter count returned identical full count matrices in every paired comparison with unmodified guide-counter 0.1.3. The performance workloads use controlled 100k/1M-read FASTQs against the public 87,437-guide Yusa library, with one or four samples.
Five paired repeats, one CPU thread. Guide-counter is faster in the tested exact-mode million-read cases. Full experimental-screen performance remains unmeasured; these performance results do not establish biological accuracy.
Install DotMatch
Published CLI 0.7.0 · Conda, containers & installation help
python3 -m pip install dotmatch==0.7.0Counting workflow
- 01 / PREPARE
Prepare inputs
Give DotMatch your library and FASTQs. Review the proposed read window before the analysis starts.
Follow the first-run tutorial - 02 / COUNT
Run assignment
Use exact, substitution-tolerant or indel-aware matching. Keep ambiguous and unmatched reads separate from unique counts.
Compare matching policies - 03 / ANALYSE
Export counts and QC
Take the raw count matrix into MAGeCK. Carry configuration, QC and methods with the result in a local review bundle.
Inspect the handoff workflow
Assignment sensitivity
Compare matching policies
In this nine-read example, exact and radius-one matching each uniquely assign three reads, but produce different per-guide counts.
Open example report| Matching rule | Unique | Ambiguous | Unmatched |
|---|---|---|---|
| Exact | 3 | 1 | 4 |
| One target within k=1 | 3 | 4 | 1 |
| Nearest target, k=1 | 5 | 2 | 1 |
One additional read has an invalid window under every policy. 5 reads change outcome between policies. Calculated with the native matcher; not a biological accuracy benchmark.
Supported assays
Barcodes & sequencing cores
Demultiplex inline barcodes and investigate collisions or unexpected unmatched reads.
Barcode diagnosticsFeature & guide capture
Assign known features and build matrices from observations with explicit cell identifiers.
Feature-matrix workflowBenchmarks and examples
Reproduce a public example, inspect count differences and compare DotMatch with your current workflow.
Yusa & Brunello CRISPR data
Recorded comparisons with MAGeCK, guide-counter and reference matchers. Runtime, memory, settings and count differences are reported together.
Read the results →Public direct-guide capture
A GSE146194 example with separate discovery and evaluation reads. Evidence for per-read guide assignment, not completed single-cell analysis.
Reproduce the case study →The example on this page
Nine synthetic reads exercise close targets, duplicate sequences, a literal N, an unmatched read and a short read. Expected assignments are checked against the native matcher.
Inspect the fixture →Pipelines and local agents
Use stable files and structured tools to prepare, preflight, run and review an assay. Inspect the installed contract before choosing a workflow.
dotmatch agent tools --json
dotmatch agent export-skill --target ./dotmatch-agentDotMatch handles known-target sequence assignment. Genome alignment, cell/UMI processing and downstream screen statistics remain separate steps. Methods & scope · Cite DotMatch
Published package 0.7.0; website source version 0.7.0.