Hi SEQanswers,
Our third MiSeq running trying our various protocol incarnations has failed, just wondered if anyone might have any ideas as to what's going wrong.
We're trying to sequence one of the human variable antigen receptor repertoires, so our amplicons are variable, containing central regions of random sequence bordered by heterologous but related sequence, that has a one of three fixed constant region on one end (which is where we target one of our primers).
Our first two goes on the MiSeq were unsuccessful; we tried to do quite a specific custom reaction, and perhaps tried too many custom variables (which was discussed briefly on this forum).
Essentially we amplified the P5 element upstream of our 3 constant sequences, and indexing and P7 oligos at the other end, then tried to sequence off 3 custom oligos directed against those constant regions. We got very few clusters, even fewer passed filter, and the sequences that came out was almost entirely low quality junk. Thinking it might be a diversity problem, we spiked in PhiX, after which we got a great run of just PhiX sequence.
After getting advice from Illumina and some contributors on here, it seemed the consensus was that either the low diversity or custom primers were likely to blame. To solve it we thought we'd introduce the Illumina SP1 site upstream of the constant primer sequence, with a random hexamer (NNNNNN) in between to provide the diversity over the first few bases. Our libraries should look like this:
P5 - SP1 - NNNNNN - PCR primer 1 - amplicon - PCR primer 2 - indexing oligo - index - P7
We then ran this on a v2 kit, in a 500bp SE reaction. We included a 5% PhiX spike in, along with two indexes (out of 12) of unrelated, randomly sheared viral genomic DNA, meaning an approximate spike in of ~21% random DNA spike in.
We don't have loads of clusters, but most of them pass filter, which was encouraging relative to our previous runs where very few did. We got about 300 bases of good quality read (with the funny increasing intensity which I'm told is typical of v2 chemistry). However, the vast majority of it is either PhiX or one of the two indexed viral samples.
The major thing that I can think of that differentiates our samples from the viral samples (apart from the diversity) is the length; the peak size of the viral samples (and the PhiX, presumably) is in the 300-400 bp range, while ours are somewhere between 700 and 900 bp. Ours also aren't normally distributed like the sheared libraries are.
I saw that some people are finding lower cluster densities on the v2 chemistry; is it possible that the shorter fragments are just out competing the longer in the bridge amplification? Or is something else going on?
As ever, any help or insight is very much appreciated.
Our third MiSeq running trying our various protocol incarnations has failed, just wondered if anyone might have any ideas as to what's going wrong.
We're trying to sequence one of the human variable antigen receptor repertoires, so our amplicons are variable, containing central regions of random sequence bordered by heterologous but related sequence, that has a one of three fixed constant region on one end (which is where we target one of our primers).
Our first two goes on the MiSeq were unsuccessful; we tried to do quite a specific custom reaction, and perhaps tried too many custom variables (which was discussed briefly on this forum).
Essentially we amplified the P5 element upstream of our 3 constant sequences, and indexing and P7 oligos at the other end, then tried to sequence off 3 custom oligos directed against those constant regions. We got very few clusters, even fewer passed filter, and the sequences that came out was almost entirely low quality junk. Thinking it might be a diversity problem, we spiked in PhiX, after which we got a great run of just PhiX sequence.
After getting advice from Illumina and some contributors on here, it seemed the consensus was that either the low diversity or custom primers were likely to blame. To solve it we thought we'd introduce the Illumina SP1 site upstream of the constant primer sequence, with a random hexamer (NNNNNN) in between to provide the diversity over the first few bases. Our libraries should look like this:
P5 - SP1 - NNNNNN - PCR primer 1 - amplicon - PCR primer 2 - indexing oligo - index - P7
We then ran this on a v2 kit, in a 500bp SE reaction. We included a 5% PhiX spike in, along with two indexes (out of 12) of unrelated, randomly sheared viral genomic DNA, meaning an approximate spike in of ~21% random DNA spike in.
We don't have loads of clusters, but most of them pass filter, which was encouraging relative to our previous runs where very few did. We got about 300 bases of good quality read (with the funny increasing intensity which I'm told is typical of v2 chemistry). However, the vast majority of it is either PhiX or one of the two indexed viral samples.
The major thing that I can think of that differentiates our samples from the viral samples (apart from the diversity) is the length; the peak size of the viral samples (and the PhiX, presumably) is in the 300-400 bp range, while ours are somewhere between 700 and 900 bp. Ours also aren't normally distributed like the sheared libraries are.
I saw that some people are finding lower cluster densities on the v2 chemistry; is it possible that the shorter fragments are just out competing the longer in the bridge amplification? Or is something else going on?
As ever, any help or insight is very much appreciated.
Comment