Design pooled sgRNA libraries for CRISPR knockout, interference (CRISPRi), activation (CRISPRa), Cas12a multiplex, base-editor, or prime-editor screens. Covers chemistry selection, on-target scoring (Rule Set 2 / Azimuth / DeepSpCas9 / CRISPRon), off-target scoring (CFD / MIT / Elevation), TSS-relative positioning, control composition, oligo synthesis layout, and post-cloning QC.
git clone https://github.com/maximilianh/crisporWebsite # CRISPOR is not on PyPI biopython pandas numpy
# Optional modern predictors
# Rule Set 2: use Broad CRISPick, or crisprScore::getAzimuthScores() in R
# (the original MicrosoftResearch/Azimuth is Python-2 only and archived) # Microsoft Research Rule Set 2 implementation
# CLI tools
# FlashFry ships as a JAR from GitHub Releases: java -jar FlashFry-assembly-1.15.jar
# For Cas12a / multiplex
R -e "devtools::install_github('crisprVerse/crisprDesignData')"
Required inputs: gene list (HGNC symbols or Ensembl IDs), target genome assembly (GRCh38 / GRCm39 / project-specific), screen chemistry (Cas9 / CRISPRi / CRISPRa / Cas12a / BE / PE), and either FANTOM5 CAGE peaks (for CRISPRi/a) or coding-exon coordinates (for Cas9 KO).
Tell the AI agent what to design: - "Design a custom Brunello-style Cas9 knockout library for 250 kinases with 4 guides per gene plus 500 non-targeting controls" - "Design a CRISPRi library against 1,200 lncRNAs using Dolcetto positioning rules against the FANTOM5 TSS" - "Build a CRISPRa Calabrese-style library targeting -150 to -75 of TSS for 800 transcription factors" - "Design an enAsCas12a paralog library covering 600 paralog pairs as 4-guide in4mer arrays" - "Design a base-editor library tiling editing windows across all exons of BRCA1 and BRCA2 for variant scanning" - "Diagnose why my freshly cloned plasmid pool has Gini 0.28 and ~3% zero-count guides"
"Design a focused Cas9 KO library targeting 350 DNA damage response genes. Use Rule Set 2 / Azimuth on-target scoring and CFD off-target filtering. Target the first 5-65% of each protein, exclude guides with GC outside 30-70% or poly-T runs, 4 guides per gene, 500 non-targeting controls, 50 AAVS1 controls, 50 CEGv2 reference essentials, 50 NEGv1 non-essentials."
"I'm screening in HCT116 cells. Should I use DeepSpCas9 instead of Azimuth Rule Set 2 for guide ranking? Compare the two and recommend."
"I have a custom gene list of 80 paralog pairs. Should I use Cas9 single-KO or Cas12a multiplex, and why?"
"Design a Dolcetto-style CRISPRi library for 600 transcription factors. Resolve TSS from FANTOM5 highest-rank CAGE peak per gene; if FANTOM5 lacks a peak, fall back to Ensembl canonical TSS and flag those genes for review. Search the Dolcetto window (-50 to +300 from TSS), ranking toward the +25 to +75 optimum. 6 guides per gene plus 1,000 NTCs."
"Compare Calabrese vs Horlbeck CRISPRa TSS targeting windows for my activation screen in iPSC-derived neurons. We need to detect modest fold-change activation."
"My CRISPRi screen has weak dropout signal even on RPL/RPS genes. Audit guide positioning against current FANTOM5 / matched neuron CAGE data and re-design any guides outside ±100 from the empirical TSS."
"Build an Inzolia-style enAsCas12a 4-guide array library covering 400 paralog pairs in the receptor tyrosine kinase family. Include singleton controls (gene A alone, gene B alone, double-NTC) so we can score genetic interaction = double_KO_LFC - sum(single_KO_LFC)."
"Design an in4mer triple-KO library testing all triplets within a 25-gene synthetic-lethality hypothesis."
"Design a CBE saturation library across BRCA1 exons 1-23 tiling every NGG-adjacent spacer that places at least one C in editing positions 4-8. Flag bystander Cs and annotate predicted amino acid change."
"Build a prime-editor library to install 320 specific ClinVar variants in MLH1, MSH2, MSH6, PMS2. Use PRIDICT2 to score pegRNA candidates and select the top-3 scored pegRNAs per intended edit."
"My plasmid pool shows Gini 0.28, skew ratio 6, and 2.5% zero-count guides on Brunello-style sequencing. Diagnose the cause (PCR bias, synthesis defect, cloning bottleneck, library age) and recommend remediation."
"We see ERBB2 dropping out as 'essential' in HER2-amplified SK-BR-3 cells. Is this a library-design issue or a copy-number bias problem?"
| Library | Modality | Guides/Gene | Coverage Need | Typical Use |
|---|---|---|---|---|
| Brunello | Cas9 KO | 4 | 500x | Modern KO standard |
| TKOv3 | Cas9 KO | 4 | 500x | BAGEL2-calibrated |
| Avana | Cas9 KO | up to 6 as published; DepMap screens a 4-guide subset | 500x | DepMap CRISPR gene-effect releases |
| Dolcetto | CRISPRi | 6 | 500x | Knockdown of essential/cuttable-toxicity genes |
| Calabrese | CRISPRa | 6 | 500x | GoF activation |
| Inzolia | Cas12a array | 4-array | 500x array | Paralog buffering / GI screens |
| Control type | Count in 70k library | Function |
|---|---|---|
| Non-targeting (NTC) | 500-1,000 (~1%) | Primary null distribution |
| Safe-harbor (AAVS1) | 50-100 | Cas9 cut-toxicity baseline |
| Olfactory receptors | 50-100 | Orthogonal non-expressed null |
| CEGv2 reference essentials | 50-100 | Positive control (BAGEL2 calibration) |
| NEGv1 reference non-essentials | 50-100 | Negative control (BAGEL2 calibration) |