Traceable Claims in Genomics

Visual-first: recommended target map (conceptual)

For gene disruption (knockout) we prioritize guides targeting constitutive early coding exons shared by all RefSeq/NM transcripts (to induce frameshifts); for CRISPRi/CRISPRa target promoters or TSS-proximal regions. The next steps produce exact sequences after you choose the reference genome build (hg38 vs GRCh38.p13 / T2T-CHM13 if desired).

Design summary (what I will deliver if you run the analysis)

• 4–6 candidate high‑scoring SpCas9 (NGG PAM) sgRNAs per gene (LMNB1 and NCL) targeting early constitutive coding exons, ranked by predicted on‑target score and off‑target risk.
• Comprehensive off‑target lists (genomic coordinates, mismatches, CFD or MIT score), filtered to <=3 mismatches genome‑wide and annotated with gene context (exonic / intronic / intergenic) and potential functional risk.
• Sequence context, GC%, predicted on‑target score (e.g., CRISPRscan/Doench 2016 or CRISPRater-style), and suggested cloning oligos for U6-driven sgRNA expression (with overhangs for popular vectors such as pX330/pSpCas9(BB)-2A-GFP or lentiCRISPR v2).
• Primers for PCR amplicons (200–400 bp) flanking each cut site for TIDE/ICE and deep sequencing QC, and a recommended experimental prioritization (3 top guides each gene) with confidence notes.
• Report with limitations, blindspots, and recommended wet‑lab validation cascade (T7E1/TIDE/amplicon NGS; single‑cell/clonal validation for frameshift verification).

Why these genes matter (selected citations)

LMNB1 (nuclear lamin B1) and NCL (nucleolin) are experimentally and clinically relevant: LMNB1 is implicated in adult‑onset leukodystrophy via duplication and expression changes, and LMNB1 has been targeted in multimodal epigenetic mapping studies; NCL is a multifunctional nucleolar protein with roles in cancer immunity and is a therapeutic biomarker candidate.

Key supporting sources:

Design principles & constraints (evidence‑based)

• Target constitutive early exons (to maximize chance of functional KO across isoforms) and avoid SNP‑dense regions if working with samples of diverse ancestry [use dbSNP / gnomAD to check].
• Use SpCas9 NGG PAM guides as first pass (broadest tool support); consider alternative nucleases (SaCas9, Cas12a) if PAM constraints or off‑target landscape demands it.
• Rank candidates by predicted activity (e.g., Doench 2016 on‑target) and penalize guides with high off‑target CFD/MIT scores; exclude guides with >1 perfect genomic match (non‑unique) or with predicted off‑targets in coding exons with <=2 mismatches unless unavoidable.
• Check guide GC% (optimal ~40–70% for stability) and avoid polyT stretches (terminate Pol III U6 transcription) and cryptic Pol III motifs.

Practical recommended pipeline (in silico -> wet lab)

1. Confirm reference sequence and genome build (hg38 / GRCh38 or T2T if desired). Retrieve full gene sequences + exon structure from RefSeq/Ensembl and list constitutive exons.
2. Generate candidate sgRNAs across target exons using CRISPOR/CRISPRscan/CHOPCHOP/FindCRISPRs; collect on‑target scores (Doench 2016), GC%, and guide position relative to ATG and splice junctions.
3. Run off‑target searches with genome‑wide alignment allowing up to 3 mismatches; compute CFD (Cutting Frequency Determination) or MIT off‑target scores and annotate off‑target loci with gene/transcript context and coding consequence predictions.
4. Filter: remove guides with predicted high‑risk off‑targets (CFD high) in coding exons; prioritize guides with high on‑target and low off‑target; keep a backup set (4–6 candidates/gene).
5. Design cloning oligos for U6 expression vectors (specify overhangs), plus PCR primers flanking cut site (amplicon 250–350 bp) for indel analysis and deep sequencing.
6. Perform wet‑lab testing tiered: transient transfection → TIDE/ICE/T7E1 screen → deep amplicon sequencing for top 3 guides/gene → clone and Sanger sequence for validated frameshift alleles if clonal KO required.

Off‑target analysis: detailed outputs you will get

• For each guide: table of all genomic off‑targets with <=3 mismatches: chromosome, coordinate (hg38), strand, protospacer sequence, mismatches positions, gene symbol (if within gene), exon/intron/intergenic annotation, distance to nearest TSS, predicted consequence (if in exon), CFD/MIT score, and presence in common SNPs (dbSNP/gnomAD).
• Summary heatmap (Plotly) of mismatch counts distribution and a prioritized risk rank for easy selection.
• Recommendation: avoid guides with any predicted off‑target that is exonic and has ≤2 mismatches to avoid unintended coding edits; consider allele‑specific checks if you have sample genotype data.

Limitations, blindspots and caveats

• In silico scores are predictive, not definitive — on‑target efficacy and real off‑target cleavage must be validated experimentally (TIDE/ICE, amplicon NGS) [predictive models have false positives/negatives].
• Reference genome differences and unmodeled SNPs/indels in your cell line or sample can create or remove off‑targets — genotype your target cell line or patient sample when possible (gnomAD/dbSNP checks mitigate population-level risks but not private variants).
• Base editors and prime editors have distinct PAM/positioning rules; this design guide focuses on nuclease (SpCas9) cutting; specify if you want base editing designs instead.
• Computational off‑target tools vary in scoring approaches (CFD vs MIT vs empirical datasets); I will report scores from multiple methods on request to reduce tool bias.

Immediate next step — run a complete, reproducible in‑silico design

I can run the full pipeline (retrieve hg38 sequences, produce candidate sgRNA lists, run genome‑wide off‑target searches, generate Tables + downloadable CSVs, cloning oligos, and PCR primers). This requires programmatic access to genome indices and off‑target tools: if you want this executed now, click the AI Scientist button to start an automated bioinformatics agent that will produce exact guide sequences and off‑target tables.

If you run the agent please specify: genome build (hg38 vs GRCh38 vs T2T-CHM13), SpCas9 or alternate nuclease, desired number of candidate guides per gene, and whether you have cell‑line genotype VCF for allele‑aware filtering.