Whole Plasmid Sequencing Service
- Scientific rigor and peace of mind.
- E. coli and other hosts will go to great lengths to avoid expressing your leaky toxic gene, including modifying your plasmid in unexpected ways that are invisible to targeted Sanger sequencing.
- Plasmid inserts are getting longer and more complex. Instead of multiple Sanger runs or synthesizing a sequencing primer or doing primer walking, sequence the whole plasmid.
- Long reads are ideal for resolving repetitive regions that stymie Sanger sequencing.
- Are you sure your plasmid isn't a dimer? Are you sure there aren't multiple plasmids in your strain? Sanger sequencing won't tell you, and we see it all the time.
- It's neither much more expensive nor slower.
In the vast majority of cases, we deliver plasmid sequencing results:
- the next morning for US customers who ship overnight using our Dropbox or UPS label generated during order submission.
- within 2-3 business days for international customers who ship overnight.
- within one business day of receipt of samples for customers who choose alternative shipping options.
Please see How to ship whole plasmid and linear/amplicon sequencing samples to get the fastest turnaround time?
Our service for plasmids is intended for a clonal population of molecules. You can send mixtures of molecular species, but since we can't predict the analysis outcome, it's at your own risk.
- If your species are very similar (e.g. differ by only a few nucleotides), the pipeline will most likely create a single .fasta consensus file with low confidence positions at SNP/indel locations. You can view those locations in your provided stats.csv and .fastq files.
- If your species are sufficiently distinct (e.g. vastly different in size or sequence), the pipeline will first attempt to make a .fasta consensus file from the highest abundance species. It will also attempt to make a consensus of other species with read counts that are >20% of that of the most abundant species. Ultimately, which species end up producing a consensus will vary depending on overall sample quality, coverage, and relative abundance/degradation of each species.
Sequencing is considered successful if the pipeline is able to generate any consensus, even if it is not your target. Re-sequencing mixtures won't change the relative proportions of the species, but you can submit multiple aliquots if you need higher total coverage.
If the pipeline does not produce a consensus for your target, you can download the raw reads from your Dashboard and bin them yourself.
If you are interested in sequencing a known mixture (e.g. barcode or variant libraries), please refer to the section below on Custom Sequencing Service.
As per Oxford
Nanopore’s specs for the chemistry
and flowcells we currently use for plasmid sequencing, the raw read accuracy is
>99%.
Deeper coverage — meaning more reads from which to build a consensus —
generally increases the accuracy of results.
The most common error modes for Oxford Nanopore are deletions in homopolymer stretches and errors at the Dam methylation
site GATC.
This limitation is expected to improve with future updates
to
their sequencing chemistry and basecalling software.
We do not guarantee any specific level of coverage.
The number of raw reads generated can vary substantially
depending
on sample quality.
Successful samples sent at the required concentration typically yield in the high dozens to
hundreds
(or thousands!)
of raw sequencing reads.
Final coverage of the consensus depends on how many of the raw sequencing reads are full-length
plasmids
and
whether any degraded plasmid reads can be aligned to the full-length consensus.
Average coverage is reported in the header line of the fasta file. Coverage over ~20x
indicates a very accurate consensus.
- .fasta file (for consensus data): We return a polished consensus sequence of each plasmid in fasta format.
- [OPTIONAL] .fastq file (for raw reads): If you like, you can elect during the order process to receive the raw fastq sequencing reads. You can also download the raw reads anytime from your Dashboard.
- .svg file (for raw reads): We return raw read length histograms for each plasmid, which provide unique insight into the contents of your samples. See details on interpreting your histograms.
- .html pLannotate map (for consensus data): We return a plasmid map for each sample, generated with the excellent pLannotate tool from the Barrick Lab.
- .gbk GenBank file (for consensus data): We return the same pLannotate map in GenBank file format.
Finally, we return two files that show how confident we are in our basecall at each position of our consensus sequence:
- .csv file (comparing raw reads to consensus data): We align the raw reads to the consensus results and return a stats.csv file that lists how well the raw reads agree at each position.
- .fastq file (comparing raw reads to consensus data): We create this .fastq file
from
the stats.csv file to visualize our consensus confidence. It contains the consensus base call
and a
“consensus score” (an internal metric of our confidence for the consensus basecall) for each
position.
This .fastq file can be viewed in software such as SnapGene Viewer to quickly identify
low-confidence
positions.
Shorter bars correspond to lower confidence positions. - IMPORTANT: Our "consensus score" is measured on the same Phred scale as standard Q-scores and is reported in the standard range of 1 to 42. This score does not correlate linearly with the percent of raw reads agreeing with the consensus. It is weighted by the number of raw reads available, ONT's machine learning algorithms for basecalling, and other statistical considerations. For example, a "consensus score" lower than 20 corresponds to a low-confidence position where under 70% of the raw reads match the consensus.
- IMPORTANT: This .fastq file for the consensus is DISTINCT from the .fastq files for the raw read data. This consensus .fastq file will be delivered with your standard sequencing results, whereas the .fastq file for the raw read data must be requested during order submission or downloaded separately from the Dashboard.
Note we do not return chromatograms (.ab1 files) because we do not do Sanger sequencing. The confidence files described above provide similar information.
The histogram displays raw read lengths between 1-25kb (for standard plasmids) and is subdivided into 100 bins. Before sequencing your plasmids, we linearize them so that we get mostly full-length sequence reads. As a result, the lengths of our raw sequencing reads reflect the lengths of the molecular species in your sample.
Ideally, your target plasmid will be the only species in the sample, and we will see one dominant peak in the read length histogram:
(Please note that even a single apparent peak MAY contain multiple plasmids of the same size, or multiple plasmids of different lengths that happen fall into the same histogram bin. Sequences that are very similar are assumed by the analysis pipeline to be variations of a single species and it will attempt to make a single consensus (with potentially low confidence positions reported in the stats.txt file); if the sequences are very distinct, it will attempt to produce multiple consensus sequences.)
If your raw reads contain varying numbers of indels (common for noisy raw reads), this may sometimes cause the read lengths to straddle a bin boundary and artifactually create an appearance of two separate peaks:
(Please note that a peak straddling a bin boundary MAY contain multiple plasmids of the same size, or multiple plasmids of different lengths that happen fall into two adjacent histogram bins. Sequences that are very similar are assumed by the analysis pipeline to be variations of a single species and it will attempt to make a single consensus (with potentially low confidence positions reported in the stats.txt file); if the sequences are very distinct, it will attempt to produce multiple consensus sequences.)
More often than you would expect, though, we see multiple peaks corresponding to multiple plasmids, or a peak of a different size than the customer expected:
(The analysis pipeline will provide consensus sequences for any plasmids in the mixture that obtain enough coverage and are at least 20% the height of the max peak in the sample. If you observe a histogram peak here that does not have an accompanying consensus delivered in your data, then coverage of that species was too low to assemble or that peak was actually a mixture of plasmid species. If sequencing fails entirely due to low coverage for all species in the mixture, you will not receive any consensus sequences, but you may find some useful information in the raw reads.)
Occasionally we see a sample with a dominant peak in addition to an abundance of degraded DNA (genomic and/or plasmid). In some cases the dominant peak may still produce a consensus, if read coverage and accuracy are sufficient:
Sometimes we see a decent number of reads for the sample but there is NO dominant peak, indicating an abundance of degraded DNA (genomic and/or plasmid) from a poor plasmid prep, or that the strain contains no plasmids:
Often, the read count is too low to distinguish any peaks or to generate any consensus:
For plasmids, “failure” refers to the failure of your sample to produce a consensus sequence with at least 10x coverage.
Our low sequencing prices and fast turnaround times do not include extensive QC to determine why plasmid samples fail. Although we do not provide definitive reasons for failure, by far the most common reasons are:
- Samples are not prepared at required
DNA
concentration.
The most common cause of this is using a Nanodrop to quantify DNA concentration. We strongly recommend using a Qubit or equivalent.
You may see evidence of this failure mode in low read counts reported in the raw read length histogram. (Please note that if you decide to evaluate the data from such samples with low read numbers, noise may exceed true sequencing signal, so we advise caution.) - Samples contain a mixture
of
plasmid species
and/or fragmented
genomic DNA or
fragmented
plasmids.
You may see evidence of this failure mode in a wide range of read lengths reported in the raw read length histogram.
To achieve optimal sequencing results, please follow our recommended sample prep and QC steps.
It is relatively rare that we cannot return a plasmid sequence, but some rate of failure is unavoidable. We may attempt to re-sequence failed samples if your sample quality and quantity permits (with follow-up results delivered in 2-3 business days). If the sample fails a second time, we will conclude that something about the sample makes it unsequenceable. Check out our Sequencing Workflow. Unfortunately we must still charge for failed samples, since we spend more time and resources on them than we do on successes.