NanoPlasmiQC: Full plasmid sequencing with ONT long-reads and automatic data analysis

The paper introduces NanoPlasmiQC, a cost-effective, automated workflow utilizing Oxford Nanopore long-read sequencing to sequence and analyze full plasmids within a single day, offering a superior alternative to traditional Sanger sequencing.

The Gist

Imagine you are a chef trying to bake a very specific, complex cake. In the past, to make sure you got the recipe right, you had to taste just a tiny crumb from the top, then another from the middle, and another from the bottom. You'd have to do this dozens of times, writing down notes for each bite, hoping you didn't miss a weird ingredient hidden in the middle. This was like Sanger sequencing: accurate, but slow, expensive, and it only let you taste tiny bits of the cake at a time.

Now, imagine you have a magical, super-long straw that can drink the entire cake in one single gulp. You get the whole recipe, from the frosting to the bottom layer, in one go. This is what NanoPlasmiQC does for scientists.

Here is the story of this new tool, explained simply:

The Problem: The "Recipe Card" Mix-up

In biology, scientists use tiny rings of DNA called plasmids as recipe cards to build new traits in plants or create medicines. Sometimes, when they copy these cards, a typo happens (a mutation).

• The Old Way: To check for typos, they used the "tiny crumb" method (Sanger sequencing). If a plasmid was large or had tricky repeating patterns, they might miss a mistake. It was also expensive; checking one plasmid could cost as much as a fancy dinner.
• The New Way: Scientists have a new technology called Oxford Nanopore (ONT). It's like that super-long straw. It can read the entire plasmid in one single, continuous read.

The Solution: The "All-in-One" Smoothie Bar

The authors of this paper, a team from the University of Bonn, created a workflow called NanoPlasmiQC. Think of it as a high-tech smoothie bar.

1. The Ingredients (Sample Prep): Instead of making one smoothie at a time, they take a tiny drop of many different plasmids (the "ingredients") and mix them all into one big bucket.
2. The Machine (Sequencing): They pour this bucket into a machine that reads the DNA. Because the machine is so sensitive, it can read the whole "recipe" of every single plasmid in the mix, even though they are all swimming together.
3. The Magic Filter (The Software): This is the real genius part. Once the machine spits out the data, a custom computer program (written in Python) acts like a super-smart librarian.
   • It takes the giant pile of mixed-up DNA reads.
   • It sorts them out, putting the reads for "Plasmid A" in one pile and "Plasmid B" in another.
   • It compares the new reads against the "expected recipe" (what the scientist thought they made).
   • It highlights any typos, missing ingredients, or weird mutations.

Why is this a Big Deal?

• Speed: In the past, checking a plasmid might take days or weeks. With this new method, you can mix, sequence, and analyze the whole batch in a single day.
• Cost: Because they can mix dozens of plasmids into one run, the cost per plasmid drops dramatically. It can be cheaper than ordering a single "crumb" test from a commercial lab.
• Accuracy: Because they read the whole thing at once, they don't miss the tricky parts where the DNA folds over itself (inverted repeats), which often hide mutations.

The "Proof of Concept"

To show it works, they tested it on a known plasmid called pBF3038.

• They ran it through their "smoothie bar."
• The computer sorted the data and found that, yes, the plasmid was mostly correct, but it also spotted a few tiny typos (mutations) that the original creators hadn't noticed.
• They even generated a visual map (like a blueprint) showing exactly where the DNA strands lined up, proving the whole thing was assembled correctly.

The Bottom Line

NanoPlasmiQC is like upgrading from a magnifying glass to a high-speed scanner. It allows scientists to check their DNA "recipe cards" faster, cheaper, and more thoroughly than ever before. This means they can spend less time worrying about typos in their data and more time actually creating new plants and medicines.

In short: It's a free, automatic, one-day system that lets you read the entire DNA instruction manual of a plasmid in one go, even if you mix a hundred of them together in a bucket.

Technical Summary

1. Problem Statement

• Limitations of Sanger Sequencing: Traditional plasmid validation relies on Sanger sequencing, which has a read length limit of approximately 1 kb. This necessitates multiple reactions and primer designs to cover full plasmids (typically 5–20 kb), making the process labor-intensive and prone to missing mutations in complex regions like inverted tandem repeats (which may harbor up to 40% unexpected mutations).
• Cost and Scalability: Commercial full plasmid sequencing services often exceed €10 per plasmid. For projects involving the generation and validation of numerous plasmids, these costs accumulate significantly.
• Historical Limitations of Long-Reads: While Oxford Nanopore Technologies (ONT) offers long reads capable of spanning entire plasmids, previous iterations were discouraged for plasmid validation due to high error rates (indels) in raw reads.
• Need for Automation: Existing workflows for plasmid analysis often require manual intervention or lack a streamlined, automated pipeline specifically designed for validating constructed plasmids against expected sequences.

2. Methodology

The authors developed NanoPlasmiQC, a cost-effective, end-to-end workflow combining wet-lab sample preparation with an automated Python-based bioinformatics pipeline.

A. Wet-Lab Protocol (Sample Preparation & Sequencing)

• Pooling Strategy: Plasmids are pooled by taking 1 µL of each sample (regardless of size differences) to simplify handling. The mixture is diluted to 10–15 ng/µL, with a total of 150 ng used for library preparation.
• Library Preparation: Utilizes the ONT Rapid Sequencing Kit (SQK-RAD114).
• Sequencing Platform: Performed on PromethION flow cells (R10 chemistry). Crucially, the authors utilize re-used flow cells (washed via a DNase-based step, EXP-WSH004) following plant genome projects, significantly reducing costs.
• Basecalling: Performed using Dorado v1.4.0 in High Accuracy (HAC) mode.

B. Bioinformatics Pipeline (NanoPlasmiQC)

The analysis is wrapped in a Python script available on GitHub, automating the following steps:

1. Reference Preparation: Cleaning of expected plasmid FASTA headers to prevent tool errors.
2. Mapping: Reads are aligned to expected sequences using Minimap2 (v2.26) with the secondary=no flag to ensure specific mapping.
3. Data Splitting: SAM files are converted to BAM, sorted, and split by plasmid reference using Samtools (v1.19) to allow independent analysis per plasmid.
4. Read Filtering: Reads are extracted and checked for duplicates using Seqkit (v2.3.0).
5. Subsampling: Due to high coverage, reads are subsampled to a target depth (aiming for >100x) to optimize assembly efficiency.
6. Variant Calling: Performed using Bcftools (v1.19) with specific quality filters (-Q 10, -q 10, QUAL>20, DP>coverage cutoff).
7. De Novo Assembly: Subsampled reads are assembled per plasmid using Miniasm (v0.3).
8. Polishing: Assembled sequences are polished using Racon (v1.5.0).
9. Visualization: Results are visualized via Integrative Genomics Viewer (IGV) for manual inspection.

3. Key Contributions

• Cost Reduction: The workflow demonstrates that full plasmid sequencing can be cheaper than a single Sanger reaction when pooling multiple plasmids, especially by utilizing re-used flow cells.
• Automation: Provides a fully automated, user-friendly Python pipeline that processes raw data to validated plasmid sequences within a day, removing the need for manual primer design or complex primer walking.
• Validation of Long-Read Accuracy: Proves that with modern ONT chemistry (R10) and high-accuracy basecalling, long-reads are now sufficiently accurate for full plasmid validation, overcoming historical concerns about indel errors.
• Open Source: The entire workflow is open-source and accessible via GitHub (bpucker/NanoPlasmiQC).

4. Results

• Sequencing Output: Two test runs on re-used PromethION flow cells yielded 12.75 Gbp and 0.58 Gbp of data, respectively.
• Read Characteristics: The N50 read lengths were 7.2 kb and 2.6 kb, respectively. While lower than typical plant genome runs, these lengths are sufficient to cover most plasmids (5–20 kb) in single reads.
• Coverage: The authors achieved sufficient coverage (targeting >100x) to validate sequences despite the pooling bias.
• Proof of Concept: The workflow was tested on plasmid pBF3038.
  • Mapping: Showed complete coverage of the plasmid length in single reads.
  • Variant Detection: Successfully identified point mutations in the sequenced copy compared to the reference.
  • Assembly: The de novo assembly successfully reconstructed the plasmid map, validating all key genetic elements.

5. Significance

• Efficiency: Enables life scientists to validate full plasmid constructs (including complex repeats and large inserts) in a single day without designing sequencing primers.
• Economic Impact: Drastically lowers the barrier for high-throughput plasmid validation in biotechnology and synthetic biology projects, making it accessible for labs with limited budgets.
• Reliability: Establishes that ONT long-reads, when combined with the NanoPlasmiQC pipeline, provide a robust alternative to Sanger sequencing, capable of detecting unexpected mutations that short-read methods might miss or require multiple reactions to find.
• Sustainability: By utilizing re-used flow cells, the workflow promotes sustainable use of sequencing resources.