PlasmidLab / Guides / Check a sequencing result
How to check a plasmid sequencing result
The facility sends back one or more reads. The question is whether the plasmid in your freezer is the one you designed, and the answer is rarely a plain yes.
What comes back
A Sanger run returns a .ab1 file per read, or sometimes .scf. The file holds the chromatogram, the base calls made from it, and a quality score per base. A .seq or .fasta next to it is the same base calls as text, with the peaks and the scores thrown away, so keep the .ab1 if you can: the peaks are what settles an ambiguous position.
One read covers roughly 700 to 900 usable bases, and the first 30 to 50 are poor, so a 5 kb plasmid needs several primers spaced along it. Whole-plasmid nanopore services return a consensus FASTA instead, which you can check the same way.
Check the reads against your design
- Open the plasmid you designed, the one the clone is supposed to be.
- Choose Analyze > Verify Clone with Sequencing Reads… and add the read files. Traces and plain sequence files can be mixed in one run.
- Read the verdict line first. It gives the coverage, then every difference grouped by what it does to the protein, then any read that could not be used.
Each read is trimmed to its good bases before anything else, at Q20 by default, which is a one-in-a-hundred error rate. A read whose ends are noisy keeps its middle; a read with nothing above the threshold is reported as unusable rather than quietly dropped.
Reads that overlap are merged, so the coverage track shows where two reads agree, where only one read speaks, and where nothing covers the plasmid at all. A difference seen by two reads is marked confirmed; one seen by a single read, or contradicted by another read, is marked so you know not to act on it yet.
Reading the differences
A difference matters according to what it does to the coding sequence it falls in, and they are listed mildest first:
- Silent, the codon changed but the amino acid did not. Common, and usually from the synthesis rather than the cloning.
- Non-coding, outside any CDS. Still worth a look if it lands in a promoter, a ribosome binding site or the origin.
- Missense, one amino acid for another, reported with the residue number so you can judge it.
- In-frame insertion or deletion, a whole codon gained or lost. The protein survives, shorter or longer by one residue.
- Nonsense, start lost, stop lost, frameshift. These end the experiment for that colony. A single base inserted or deleted inside a CDS is a frameshift, and it is the most common way a clone that looks right on a gel is dead.
Each coding sequence also gets its own line: intact, changed, broken, or only partly read. A gene reported as intact over 100% of its length is the sentence you were looking for, and it is not the same as the plasmid being perfect, because the backbone may be uncovered.
When the trace says the colony is mixed
A clean read has a single peak at each position, with a secondary peak under about 1% of positions. When a second peak reaches 30% of the called one across many positions, two templates went into the reaction: usually two plasmids in the same colony, sometimes a primer binding in two places. The check reports the percentage and names the read.
This is the failure that the base calls alone hide. The caller reports whichever template is stronger, so the text looks like a clean match while half the plasmid in your tube is something else. Restreak, pick single colonies and sequence again.
Gaps, and what a pass really means
The uncovered ranges are listed by position. A region nothing read is a region you know nothing about, which matters most at the junctions you made: those are where cloning goes wrong, so a read should cross each one. Where the reads are short of the junction, order a primer that sits 50 or more bases before it.
If several reads belong to one region and you want the consensus first, Analyze > Assemble Sequencing Reads… builds the contig, and Analyze > Align Against This Sequence > Sequences from Files… lines up reads without the verdict.