three questions worth answering
The best summer projects are those where the student can answer one of these by week 12:
- Can I predict ThermoCas9 activity from local methylation state?
- Can I identify candidate methylation-selective target sites in a cancer type?
- Can I measure whether methylation-sensitive discrimination holds in a simplified assay?
The 2026 Nature result is that ThermoCas9 discrimination is PAM-methylation-dependent. The current evidence base is preclinical and cell-focused, so a summer project should stay at the level of assay development, computational target discovery, or small-scale validation — not animal work or trial design.
Best-fit undergraduate project scopes
Computational target discovery
The safest and highest-probability undergraduate project.
- Question
- Can we build a ranked list of ThermoCas9-compatible loci that are likely to be tumor-selective because the PAM site is hypomethylated in tumor and methylated in normal tissue?
- What the student does
- Use public methylation datasets (TCGA, ENCODE, GEO) or a lab-provided dataset, scan for ThermoCas9 PAM-like sites (
5'-NNNNCGA-3'/5'-NNNNCCA-3'), overlay methylation at the key PAM cytosine, and rank candidates by tumor-normal separation. - Deliverable by end of summer
- A shortlist of 10 to 50 candidate loci in one cancer type, with a scoring framework and figures.
- Why it fits
- Directly tests the main translational premise that local PAM methylation, not bulk methylation, should drive selectivity. Inference follows from the PAM-centric mechanism in the Nature study.
- Skills learned
- Python or R, methylation data analysis, sequence scanning, genomics visualization, basic translational biomarker logic.
- Full project plan
- Computational discovery of methylation-selective ThermoCas9 target sites in cancer — 10-week deep-dive: 5 phases, weekly schedule, mentor checklist, figure plan, technical limitations.
- Publication-path upgrade
- A methylome-guided framework for identifying tumor-selective ThermoCas9 target sites — how to reframe the same project as a publishable methods paper: feature-based scoring, uncertainty modeling, ranking benchmarks, three manuscript scope tiers, reviewer Q&A.
Cell-free cleavage assay with methylated vs unmethylated DNA
The best wet-lab undergraduate project if the lab already has reagents and a CRISPR setup.
- Question
- Does ThermoCas9 show differential cleavage between matched methylated and unmethylated target substrates at a candidate PAM?
- What the student does
- Generate or obtain synthetic DNA targets with the same sequence except for methylation state, incubate with ThermoCas9 RNP, and quantify cleavage by gel or capillary assay.
- Deliverable by end of summer
- A small matrix of targets showing whether methylation at the PAM blocks cleavage and whether the effect varies by sequence context.
- Why it fits
- Mechanistically sharp, low-cost compared with cell work, and aligned with the published finding that methylation at the PAM-bearing cytosine suppresses activity while methylation in the protospacer matters less.
- Risk
- Only feasible if the lab already has ThermoCas9 access, an RNP workflow, and methylated substrates.
Literature-plus-data mini-review with target nomination
Works well for an undergraduate new to the field.
- Question
- Which cancer types and loci look most plausible for methylation-selective ThermoCas9 development?
- What the student does
- Review the ThermoCas9 literature, summarize the mechanism, compare it to other methylation-sensitive CRISPR systems (notably AceCas9), and integrate public methylation datasets to nominate one disease area.
- Deliverable by end of summer
- A review-style report plus one data-driven target nomination figure set.
- Why it fits
- Produces something publishable internally and teaches the student how to connect mechanism to translational strategy.
Best-fit graduate student project scopes
PAM-site methylation as a predictive biomarker
- Question
- Is editing efficiency across loci better predicted by PAM-site methylation than by global methylation or chromatin proxies?
- Scope
- Select a panel of candidate loci across one or two cell lines, quantify local methylation, perform ThermoCas9 editing, and model predictors of activity.
- Deliverable
- A small predictive framework showing which variables explain editing best.
- Why it's strong
- Addresses the central translational bottleneck: patient selection. Current evidence suggests local PAM methylation should be the dominant variable, but that needs broader validation. Realistic 10–12 week graduate rotation if the cell-editing pipeline already exists.
Tumor heterogeneity pilot
- Question
- How much does intratumoral methylation heterogeneity erode expected ThermoCas9 selectivity?
- Scope
- Use existing methylation datasets, ideally paired bulk and single-cell or spatial data, to estimate how often a "good" candidate target is actually clonal versus heterogeneous.
- Deliverable
- A quantitative framework for ranking targets by expected resistance risk.
- Why it matters
- A major translational concern is that epigenetic heterogeneity could create escape subclones even when bulk tumor methylation looks favorable. Forward-looking but highly relevant given the mechanism. Purely computational — realistic for summer timing.
Delivery-format comparison in vitro
- Question
- Does transient delivery preserve methylation-selective behavior better than longer-expression formats?
- Scope
- Compare RNP versus mRNA or plasmid, at a small number of loci, in a cell model with known methylation differences.
- Deliverable
- A small but useful dataset on editing magnitude, selectivity ratio, and cell viability.
- Why it fits
- The current platform evidence points toward RNP as the practical direction for engineered ThermoCas9. Even a limited comparison is translationally relevant. Better suited to a graduate student because it can fail for many technical reasons.
Project format by experience level
| Experience | Realistic options | Best single pick |
|---|---|---|
| Undergraduate | Computational target discovery · simple in vitro cleavage assay · structured review + target nomination | Computational target discovery, optionally with one pilot validation assay |
| Graduate student | Biomarker prediction across loci · methylation heterogeneity / resistance modeling · small in vitro delivery comparison | PAM-site biomarker prediction across a small locus panel |
Too big for a summer project
- Designing a full clinical trial package
- Building a new delivery platform from scratch
- Animal efficacy studies, unless the system is already running
- Genome-wide off-target mapping as a standalone first project
- Discovering and validating a whole new ThermoCas9 variant
These are thesis-scale or team-scale efforts.
A concrete example scope
Example · graduate summer project
Can local PAM methylation predict methylation-selective ThermoCas9 editing across candidate breast cancer loci?
Aims:
- Identify 15 to 20 ThermoCas9-compatible candidate sites in breast cancer methylation data.
- Measure local methylation status in 2 cell lines (e.g., MCF-7 and MCF-10A).
- Test editing at the top 5 to 8 loci.
- Model editing versus PAM methylation level.
Success criterion: Show that PAM methylation explains a meaningful fraction of editing variability and generate a ranked target list for follow-up.
A very good undergraduate version is the same project, stopping after Aim 1 plus perhaps one pilot assay.
Indicative 10-week schedule
| Week | Computational scope (undergrad) | Biomarker scope (graduate) |
|---|---|---|
| 1 | Literature: read Roth et al. 2026; survey methylation-sensitive CRISPR landscape | Same + audit lab's editing + methylation pipeline |
| 2 | Set up Python/R env; pull TCGA / ENCODE methylation arrays for chosen tumor type | Define candidate panel (~15–20 loci) from public methylation data |
| 3 | PAM scanner: regex over reference genome, intersect with CpG/CpC sites | Order sgRNAs and primers; bisulfite-sequence panel in 2 cell lines |
| 4 | Map β-values onto PAM cytosines; tumor vs normal differential | Confirm methylation patterns; finalize top 5–8 loci for editing |
| 5 | Build ranking score; sensitivity to thresholds and cohort size | RNP transfections with WT or CE ThermoCas9; harvest at 72 h |
| 6 | Generate top-N candidate table; figure-quality plots | Indel quantification (ICE / CRISPResso2); replicate runs |
| 7 | Pilot validation: pick 1 candidate, run a simple in vitro cleavage assay if feasible | Build editing-vs-methylation regression; alternative predictors |
| 8 | Cross-check against orthogonal datasets (e.g., second cohort) | Single-cell or spatial methylome lookup for top loci (heterogeneity check) |
| 9 | Draft methods + figures; assemble GitHub repo with notebooks | Draft methods + figures; release code and ranked target list |
| 10 | Final report + symposium poster + GitHub release | BioRxiv preprint draft + symposium talk |
Publication targets
A bounded summer project can realistically produce one of the following publication outputs. List from quickest to most ambitious:
- BioRxiv preprint with code and ranked target tables — fastest, citable, and signals priority.
- Conference abstract at iGEM (for synthetic biology), AACR (oncology), ASGCT (gene therapy), or institutional summer-research symposia.
- Short paper or correspondence in CRISPR Journal, ACS Synthetic Biology, Nucleic Acids Research, Bioinformatics Advances, or NAR Genomics and Bioinformatics — for the computational and biomarker scopes especially.
- Methods paper in STAR Protocols or Bio-protocol — appropriate if the deliverable is a reusable assay or pipeline.
The computational target-discovery and PAM-biomarker scopes have the highest publication probability because they produce datasets and tools that can be released and cited.
Mentor-side prerequisites for a smooth summer
- Pre-built editing or cleavage pipeline if any wet-lab work is planned
- Access to ThermoCas9 protein or expression construct (or reagent supplier identified before week 1)
- Pre-selected cell lines with prior methylation profiling (or commitment to bisulfite-sequence in the first 3 weeks)
- Computational compute (laptop is fine for most candidate panels; cluster only needed for cohort-scale scans)
- A weekly 30-minute check-in to keep scope tight
Recommended starting picks
- Undergraduate: computational target discovery with one pilot validation assay.
- Graduate student: PAM-site biomarker prediction across a small locus panel.
Both scopes are narrow enough to finish, scientifically meaningful, and directly connected to the actual novelty of the ThermoCas9 methylation-selective work.
next outputs
This guide can be turned into a specific 10-week project plan for one student
(with week-by-week milestones, materials list, and a final presentation outline) or a
mentor-side onboarding doc. Contact
contact@thermocas9.com.
Source
Roth M.O., Shu Y., Zhao Y., Trasanidou D., Hoffman R.D., et al. Molecular basis for methylation-sensitive editing by Cas9. Nature (2026). DOI: 10.1038/s41586-026-10384-z. Open access (CC BY-NC-ND 4.0).