Peaktrace vs ABI KB basecaller - (Aug/07/2014 )
What do you think of a DNA sequencing facility that is using Peaktrace ( http://www.nucleics.com/peaktrace/ ) to generate "outstanding" chromatogram???
Usually chromatogram generate by ABI platform will show the peak become broader and resolution become lower due to limitation of the platform after 700 - 800bp eventhough most of the time it give reading up to 1000+ bp. Please attachment (SEQ1.ab1) for example.
However some facility are using Peaktrace to call "better" chromatogram which can show to end user that the chromatogram can show very nice, beautiful and high resolution peak even it is near 1000bp. Please attachment (SEQ2.ab1) for example.
Thus, there is voice saying that the facility which generate chromatogram using Peaktrace do a better job than the one that is using ABI KB basecaller.
What do you think about this??? The 2.ab1 file I attached are actually from same sequence trace data.
Hope to hear your opinions... Thanks!!!
off topic: please no double posts
Is this supposed to be an advertisment of some kind?
In that case, (apart from the fact you are a nasty spammer) I'm quite impressed.
I don't sequence many longer templates, because I know I can't get such long good quality basecalls but sometimes you need it.
I didn't have anything better now to try than this "slipped" sequence, but you can see a significant difference in the quality of the basecalling after the polyC stretch (and with around 90% accuracy, which is amazing). The rest of the sequence looks more "clear" with less background.
(it's bigger that the preview, save it to disk for full size)
They offer 6 traces a day for free. I think this may be a very interesting option, since I rarely have more "problematic" sequences. At least for now. I believe If it's really that good, you can get used to it very quickly and crave for more
But I need to note, that even in the original ABI basecaller you can improve quality of some sequences (those that have a sudden signal drop, for example) if you analyze them separately from the rest.
(BTW sequence data files are AFAIK one of the unsupported file formats on the forum, still, you can workaround this by renaming to an "acceptable" extension)
No... I am not advertise for Peaktrace.
I am asking opinions that should sequencing facility using this software to "call" better chromatogram ? Because there is argument in our lab that the other facility which don't use it do not do well in providing"long" sequencing data... But in fact if we put the non-peaktrace-called sequence trace in peaktrace, It give similar outstanding look of the chromatogram...
Now we can't decide which facility is doing better. We can't say the facility using Peaktrace doing better than the other right??? In the sense that they actually "modify" the chromatogram...
First I would say, it doesn't matter how someone does produce better sequencing data if he does.
You may improve quality of the reads using different polymer type, different capillaries, etc. That is costy. If a software can basecall the same data into better sequence, then it's usually much less costy. But sometimes, if the data is too bad, no software can save it.
It's difficult to say if the information is there and is just plainly wasted by original software, or the PeakTrace tries to "guess" more than actually improve data.. you can't tell unless you try (or unless you understand the algorithms which I don't) on long difficult sequences.
Choose the facility, that gives you better results cheaper, that's all. But you must be sure if the result is really "better".
Today I tried it on some long plasmid sequences and it improved reads significantly, however in a very bad looking "original" sequence, the polynucleotide stretches were often indistinguishable (just a huge wide peak), you couldn't tell if there are 4 A's or 5. But at least you knew, it was a bad sequence there.. PeakTrace did improve that part of sequence too, almost magically, but in several cases "miscalculated" the poly stretches, for example showing clear 4 A's, when there should be actually 5 or so.
That is not good, since it didn't look suspitiously bad to doubt the read.. withou original sequence I could thing there is a deletion, because the sequence read look convincing.
This is a part of what I pointed out as a con of these "improving" algorithms, sometimes they just guess and make things up from a seriously bad data, presenting them looking "nicer" so you wouldn't doubt it when you really should.
Make your own conclusion out of this.