Overview

RPA/CRISPR-Cas12a is a growing technique that pairs Recombinant Polymerase Amplification (RPA) with the CRISPR-Cas12a system for rapid, sensitive nucleic-acid detection (see Diagnostic Overview). Designing these assays requires coordinated RPA primers and Cas12a crRNAs. RPA primers are short synthetic DNA sequences that initiate amplification by binding to specific sites on the target DNA, enabling exponential copying of the region of interest under isothermal conditions. Cas12a crRNAs (CRISPR RNAs) are guide molecules that direct the Cas12a nuclease to a complementary DNA sequence within that amplified region, triggering the enzyme’s cleavage activity. Current design processes are not integrated, which creates hand-offs, trial-and-error, and uneven assay performance (Malci et al., 2022).

To validate the need of a new tool, we reached out to active researchers in this field.

While there are tools for RPA primer design and CRISPR crRNA selection, there is a lack of integrated software that optimizes both. A new tool could significantly enhance experimental efficiency and assay performance.

— Dr. Fengming Wang, Changzhou Center for Disease Control and Prevention (China)

Our group often orders as many as ten RPA CRISPR-Cas12a pairs at a time due to the inaccuracy of current design methods, and there is still no simple, accessible application suitable for routine use.

— Dr. Daniel Richards, deMello Lab at ETH Zurich

Combined Amplification & Spacer Engine for RPA-Cas12a (CASPER) is an integrated RPA primer and CRISPR-Cas12a crRNA design tool built to generate sets that work together effectively. Given a target DNA sequence, CASPER proposes candidates and computes scores for RPA amplicons and primers, Cas12a crRNAs.

It ranks candidates with a composite score that blends thermodynamic modeling of primer/guide hybridization and secondary structure, rule-based constraints (e.g., seed-region conservation, structural stability), background homology penalties for likely off-target activity.

CASPER was experimentally validated in two ways. First, the predictor scored four wet-lab primer pairs (1.3, 1.2, 1.4, 1.1); the predicted ranking matched experimental outcomes (1.3 > 1.2 > 1.4 > 1.1). Second, CASPER generated a de novo, integrated primer–crRNA set (1.5), which was ordered from Integrated DNA Technologies and evaluated by gel electrophoresis; the CASPER-designed primers were more effective than all others tested. The crRNA designed by CASPER also outperformed an alternative crRNA with a greater cleaved 6-FAM reporter concentration in the same assay, confirming that CASPER’s designs translate into tangible in-vitro success (see Results).

To improve usability, CASPER is available as a web app, a Python package, and a command-line interface (CLI). It accepts FASTA inputs and returns clean CSV outputs to align with common bioinformatics workflows.

CASPER is available as a web app, a Python package, and a command-line interface (CLI).
It accepts FASTA inputs and returns clean CSV outputs to align with common bioinformatics workflows.

Judging

1. How well is the software compatible with, and does it leverage, existing synthetic biology standards (e.g., SBOL, other RFCs, data formats, etc.)?
   CASPER fits into standard synthetic biology workflows by reading FASTA files (a universal format for protein and nucleotide sequences) and exporting CSV for seamless use in notebooks and other tools. While CASPER does not yet natively support SBOL, its internal objects—target locus, primer pair, PAM/protospacer, constraints, and composite score—map naturally onto the SBOL 3 data model.

2. Was this software validated by experimental work?
   Yes. CASPER was validated in two ways. First, the predictor scored four primer pairs designed with external tools and ordered by our wet lab team; CASPER’s predicted ranking matched experimental performance exactly (1.3 > 1.2 > 1.4 > 1.1) (see Results). Second, the generator produced a new, integrated primer–crRNA set that we evaluated experimentally: the CASPER-designed primer pair showed significantly higher amplification than all four comparators, and the paired crRNA outperformed our alternative crRNA in the same assay—confirming that CASPER’s designs translate into tangible in-vitro success.

3. Can the software be useful to other projects?
   Yes. Current tooling is fragmented—labs often use separate tools for RPA and CRISPR-Cas12a, leading to inefficient, trial-and-error design. Researchers we consulted emphasized the absence of a truly integrated workflow and reported sometimes needing to “order 10 pairs at a time” to find a working set. CASPER addresses this gap by co-designing primers and crRNAs in a single workflow, improving accuracy and reducing iteration (see Validation).

4. How well can the software be integrated with external tools/software applications (including APIs, packages, etc.)?
   CASPER provides a robust CLI and programmatic API for straightforward integration into external scripts and pipelines. After cloning the GitLab repository, users can run the CLI with clear flags and YAML-based configs for reproducible, batchable runs. The FastAPI backend exposes endpoints so teams can create designs, integrate with scripts/notebooks, and feed results directly into their own workflows.

5. Is the software user-friendly?
   The web app offers a polished, no-code workflow: users paste or upload FASTA, set design constraints, filter/sort candidates, and export results as CSV. Clear defaults, concise inline guidance, and responsive interactions help newcomers progress quickly from sequence input to a ranked shortlist, while advanced controls enable precise tuning for experienced researchers.

6. How well is the software written and documented for future groups to extend and improve?
   CASPER is structured for maintainability and extension. The GitLab repository includes complete documentation for installation, configuration, and usage, with worked examples. A clear pipeline description explains the full path from primer/crRNA generation to scoring and ranking, enabling developers to understand and modify the workflow confidently. Installation scripts streamline setup, and in-code comments with organized modules make it easier to build upon. Together, these features provide a strong foundation for future teams to adapt and improve CASPER over time.

Usage

When our wetlab team began designing sequences for the RPA and CRISPR-Cas12 systems, they found no tool that generated and evaluated primers and crRNAs in a user-friendly way. Existing resources were command-line based, hard to install, or lacked interfaces suitable for researchers without coding experience. We built CASPER as both a web app and a CLI so that researchers can either click through a design flow or script repeatable batch runs.

Web app

A React front end with a FastAPI back end provides a clean, no-code workflow: paste or upload FASTA/sequence, choose design constraints, and generate candidates with live scoring and side-by-side comparisons. Results can be filtered and sorted in real time, and exports (CSV) make it easy to move designs into lab notebooks and pipelines (see Figure 1). It’s optimized for fast iteration so students and experienced researchers can reach a ranked short list in minutes.

Fig. 1. CASPER Web App Running Prediction of RPA Primers and crRNAs

Command-line interface (CLI)

git clone https://gitlab.igem.org/2025/software-tools/lambert-ga.git
cd lambert-ga/CASPER/CASPER-Cli
pip install -e .

conda create --name casper-bio
conda activate casper-bio
pip install -r requirements.txt

casper --target-fasta target.fasta

casper --target-fasta target.fasta --input-csv existing_primers.csv

casper --target-fasta target.fasta

casper-generate --target-fasta target.fasta --output pairs.csv

casper-filter --input pairs.csv --output filtered.csv --min-gc 30 --max-gc 70

casper-features --input filtered.csv --output features.csv
casper-rank --input features.csv --output ranked.csv --num-sets 10

from casper import SequenceData, PairGenerator, PrimerFilter, FeatureCalculator, Ranker

# Load and generate primer pairs
sequence_data = SequenceData("target.fasta")
sequence_data.preprocess()

generator = PairGenerator(sequence_data)
generator.generate_primers_pairs(
    sequence_data.sequence,
    primer_len=[18, 25],
    amplicon_len=[70, 200],
    crrnalen=[17, 24]
)
generator.to_csv("pairs.csv")

# Filter, calculate features, and rank
filter_obj = PrimerFilter("pairs.csv")
filtered = filter_obj.run([30, 70])  # GC content range

features = FeatureCalculator("filtered.csv")
features.compute_all_features()

ranker = Ranker("features.csv")
ranker.rank()
ranker.to_csv("ranked.csv", num_sets=10)

For reproducible batch runs and automation, the CASPER CLI makes design runs simple and repeatable. Commands like casper generate, casper filter, casper score, and casper rank take FASTA/CSV inputs and a short YAML config, then return CSV outputs. Clear flags, example configs, and helpful “—help” text reduce setup to minutes. A lightweight Python API mirrors the CLI so researchers can call CASPER directly from notebooks and code.

CASPER

CASPER automatically designs RPA primers based on a target sequence, then generates a corresponding crRNA that aligns with the produced RPA amplicon. To increase accuracy and assay reliability, each primer–amplicon–crRNA set is evaluated together for compatibility (see Fig. 2).

Fig. 2. CASPER Complete Pipeline

Generation & Filtering

CASPER enumerates potential primer–crRNA sets on the target sequence. It keeps only primers whose amplicon contains a valid PAM so a matching crRNA can be designed. For each surviving set, CASPER finds all potential crRNAs downstream of the PAM and combines them into a complete candidate (forward primer, reverse primer, amplicon, PAM, protospacer). A pre-scoring filter removes weak candidates before heavy thermodynamic or structure calculations. These quick checks reduce the search space so only promising sets proceed. Default thresholds are provided and may be adjusted by researchers.

Pre-scoring filters

Ranking Framework

CASPER produces a composite ranking score for each RPA primer–crRNA set by weighting key features. We curated 100+ primer–crRNA sets from RPA–CRISPR-Cas12a literature with reported outcomes and trained an L2-regularized logistic regression to determine feature weights. The model estimates probability of success while shrinking coefficients to avoid overfitting, yielding stable, data-driven weights. The result is an empirically weighted composite score scaled 0–1.

CASPER evaluates three categories

Note: Weights are data-driven defaults and can be overridden in the YAML config for project-specific tuning.

We combine these three aspects into a single score (0 to 1) representing overall predicted assay efficiency.

Code and documentation for the scoring framework →

Evaluation

CASPER uses its scoring system not only to generate but also to evaluate existing designs.

Users can input their own target sequence, RPA primers, and crRNAs. In turn, CASPER returns both overall composite scores and feature-level partial scores, giving deeper insight into strengths and weaknesses.

Results

We evaluated CASPER in two phases: (i) in-silico benchmarking against contemporary design tools and (ii) wet-lab validation of selected primer–crRNA sets.

In Silico Results

Compared with existing options such as PrimedSherlock — which can take ~8 hours for a full design run on an advanced GPU (3900XT + 2080 Ti) and requires complex setup — CASPER delivers end-to-end designs in under 10 minutes on a single CPU while remaining accessible to non-specialists (Mann, 2025). Rather than focusing solely on target conservation, CASPER emphasizes functional effectiveness by jointly optimizing thermodynamic balance (ΔTₘ between primers), guide secondary-structure stability, and mismatch-weighted off-target risk.

In an in-silico screen of 1,000 primer–crRNA candidates, CASPER evaluated designs against three criteria that prior literature identifies as crucial for RPA–CRISPR-Cas12a performance: (1) mismatch-weighted off-target impact ≤ 0.40 to prioritize specificity, (2) primer ΔT_m < 1 °C to ensure synchronized isothermal amplification, and (3) guide secondary-structure MFE > −1.0 kcal/mol to avoid inhibitory folding. Under these pre-specified thresholds, 685/1,000 candidates generated by CASPER satisfied all three requirements, yielding a 68.5% pass rate while maintaining runtimes short enough for routine iterative design.

Mismatch-weighted off-target impact ≤ 0.40 (specificity)

Keeping the mismatch-weighted off-target impact low is essential because even a few mismatches between the crRNA and off-target DNA can produce false positives, lowering the assay’s reliability and accuracy. Tools that help predict and minimize these off-target effects greatly enhance test specificity (Zhang et al., 2022).

Primer ΔTₘ < 1 °C (synchronized amplification)

If primer melting temperatures are not closely matched, amplification can become uneven or non-specific, reducing sensitivity and leading to false negatives. Careful primer selection by in-silico design ensures robust and reliable amplification essential for diagnostic consistency (Chen et al., 2025).

Guide secondary-structure MFE > −1.0 kcal/mol (avoid inhibitory folding)

When guide RNAs form strong inhibitory structures (more negative MFE), they may not bind properly, stalling the CRISPR reaction. Screening crRNA candidates for favorable (less negative) MFE values with software helps avoid this issue, improving assay efficiency and accuracy (Li et al., 2024).

Experimental Validation

To validate CASPER’s effectiveness under experimental conditions, we incorporated two complementary testing strategies.

A) Ranking existing wetlab primers: First, we applied CASPER’s predictor model to evaluate the RPA primer pairs ordered by the Lambert iGEM wet-lab team. The team used four primer pairs (See RPA Primers here), and we compared CASPER’s predicted rankings with our team’s experimental results to determine if CASPER could accurately identify the best-performing designs (See Fig. 3).

Fig. 3. CASPER Experimental Comparison

This ranking aligned with the outcome of our wetlab experiments, confirming the accuracy of our results.

B) De novo design and experimental validation: We designed a complete primer–crRNA set using CASPER rather than external manuals, selected top-ranked candidates targeting the same region as our RPA assay, and ordered a new primer–crRNA pair (1.5) from IDT for in-vitro testing.

When utilizing primer set 1.5, we found that we were able to visualize fluorescent bands at the expected amplicon base pair length of 193 base pairs using gel electrophoresis. Bands were observed in samples with protein concentration up to 250 days post-infection, supporting the functionality of CASPER-designed primers (see Fig. 4a). The CASPER-generated primer analysis outperformed all other primers tested (See Fig. 4b).

Previous slideNext slide

Slide 0 of 0

Figure 4a. CASPER Gel Electrophoresis Results. Figure 4b. CASPER ordered primer 1.5 in comparison with greatest non-CASPER designed primer.

We also tested our generated crRNA sequence (BBa_25ZB4FN5), which was designed and ordered through CASPER. The CASPER-designed crRNA was tested alongside the RPA-amplified PDL product to evaluate its detection performance. Our results showed that the CASPER-generated crRNA induced strong collateral cleavage activity on the FQ reporter, producing higher detection efficiency and clearer fluorescence signals than the other experimentally designed crRNA (See Figure 4). Its performance was not statistically significantly different from the positive control, further supporting CASPER’s precision and reliability in guide RNA design. Although differences among experimental crRNAs were not statistically significant, the CASPER-generated crRNA consistently demonstrated superior overall performance trends.

Figure 5. Graph of Ca12a reaction comparing CASPER crRNA enzyme activity with positive and experimental reactions, demonstrating successful Cas12a activity indicated by greater cleaved 6-FAM reporter concentration than in negative controls.

References