CRISPR Guide Allele-Specificity Evaluation
A completed computational genomics study evaluating CRISPR guide discrimination between clinically reviewed variant and reference alleles, with whole-genome off-target evidence.

Project definition
Problem statement
A CRISPR guide aimed at one disease-associated allele may also bind the corresponding reference allele or similar genomic sites.
The engineering problem is to evaluate allele discrimination, sequence quality, genome-wide similarity, and rejection evidence together without treating an in silico score as proof of editing safety or efficacy.
Project objectives
- Prepare a balanced panel of forty clinically reviewed small variants from the current ClinVar release.
- Enumerate SpCas9 NGG candidates around variant and reference alleles.
- Measure variant position, PAM creation, sequence quality, and CFD residual binding.
- Search the GRCh38 reference genome with Cas-OFFinder and verify every retained hit against the source genome.
- Test whether one intentional spacer mismatch improves the calculated variant-to-reference discrimination ratio.
- Retain complete results, figures, provenance, tests, references, and interpretation limits.
Project structure
Project components
Variant preparation
Parses ClinVar records, applies review and significance rules, balances the panel, and extracts GRCh38 sequence windows.
Guide enumeration
Builds variant and reference allele windows and finds forward and reverse SpCas9 NGG candidates.
Allele scoring
Calculates sequence quality, mismatch position, PAM effects, and CFD residual binding to the reference allele.
Genome search
Runs Cas-OFFinder and verifies sequence, coordinates, strand, mismatch count, and hit summaries against GRCh38.
Experiment and evidence
Selects one representative per variant, tests intentional mismatches, and retains CSV, JSON, figure, test, and document evidence.
Methodology
Project workflow
- 01Prepare variants
Select reviewed ClinVar records and verify their GRCh38 alleles and sequence context.
- 02Generate candidates
Enumerate NGG-compatible guides for each alternate allele and record how the variant creates discrimination.
- 03Score alleles
Compare alternate and reference targets with sequence-quality and CFD calculations.
- 04Search the genome
Run and verify the bounded Cas-OFFinder search for the shortlisted guides.
- 05Select and interpret
Choose one representative per variant, test intentional mismatches, and review success and rejection cases.
Demonstration scenario
The study retains one guide for each of forty variants. Thirty-nine selected guides have no exact GRCh38 hit under the declared search, while the low-complexity IQCB1 case is retained as a documented rejection example. The intentional-mismatch sweep finds modest calculated improvements for twenty-five guides but no twofold improvement.
Engineering
Tools and method
- Tools
- The project uses Python 3.12, Cas-OFFinder 2.4.1, ClinVar, GRCh38, Pandas, Matplotlib for subject analysis, simulation, and results.
- Genomic data
- The workflow uses recorded ClinVar and GRCh38 releases with checksums and explicit coordinate handling.
- Guide analysis
- Python implements candidate enumeration, reverse-complement handling, quality rules, and CFD scoring.
- Off-target search
- Cas-OFFinder 2.4.1 performs the genome scan; a separate parser verifies every retained row against GRCh38.
- Evidence
- Pandas, JSON, and Matplotlib retain candidate, selection, hit, mismatch, summary, and figure outputs.
- Verification
- Automated tests, dependency audits, document checks, hashes, and a Linux Docker run validate the completed study.
Testing
Evaluation
Evaluation measures
- Candidate yield and variant-position class across the forty-variant panel
- Alternate-to-reference CFD residual binding for every selected guide
- Verified exact, one-mismatch, total, and maximum-CFD genome hits
- Bounded specificity index and sequence-quality tradeoffs
- Intentional-mismatch discrimination ratio across 2,400 combinations
- Automated test coverage, dependency audit, document validation, and Docker execution
Project boundaries
- The work is an in silico engineering evaluation and does not demonstrate editing efficiency, allele selectivity, or biological safety.
- ClinVar significance describes submitted variant interpretation and is not a guide-design recommendation.
- The bounded genome search, CFD model, reference assembly, PAM choice, and sequence rules define the reported result.
- Wet-laboratory validation, delivery, cell context, repair outcomes, clinical use, and therapeutic decisions are outside scope.
Included
- 01Complete allele-specific CRISPR evaluation source code
- 02Forty clinically reviewed ClinVar variants across forty genes
- 03One hundred forty-three candidate guides and forty selected representatives
- 04Verified Cas-OFFinder evidence from 226,867 raw search rows
- 05CFD allele scoring and 2,400 intentional-mismatch comparisons
- 06Twelve generated result figures and three attributed literature figures
- 07Thirty-eight automated tests with 97.12 percent combined coverage
- 08Complete project files, calculations, results, and analysis material in a private GitHub repository
- 0984-page project documentation in PDF and editable Word formats
- 1010-page setup and usage guide in PDF and editable Word formats
- 11Fifty-five annotated references
Project record
No information is collected on this page.
- Permanent project ID
- GP-BT-0WE4XKX
- Catalogued
- 21 Aug 2026
- Completed
- 26 Aug 2026
- Verified
- 26 Aug 2026
- Demonstration
- Included in repository
Handover
After purchase
- 01Payment is confirmed
The project is marked unavailable and cannot be purchased again.
- 02Repository access is granted
The buyer's submitted GitHub account receives access to the private repository.
- 03The purchase record is delivered
The certification sheet is prepared from the reviewed buyer details and sent privately by email.