Solexa adapters

[Paedhaemdoc]

Trying to understand how Solexa adapters work in multiplex sequencing...Would be great if someone can explain how the specificity of the adapters to the forward and reverse strand is maintained?

One adapter has the sequencing primer
The other adapter has the index and the index sequencing primer.

Will this determine in which direction the sequencing is performed?

Hope my query makes sense..

Paedhaemdoc

[marklar]

I'm not sure how the specificity is maintained, but as for the adapters/seq primers/indexes, have you seen this? http://www.illumina.com/downloads/SQ...0-2008-011.pdf
It's a very basic sheet on their multiplexing, but it does have some good pics showing the orientations of the primers and the "three-read sequencing strategy" (if you're doing paired-end).

[Will]

Hi there,

I am interested to find out how the adapters work in multiplexing also.

But the link has stopped working.

If it's just the standard Illumina link to multiplexing no worries.

What I really want to understand is how the third primer incorporates into the product during the reaction.

Cheers

Will

[odile]

Adapter sequences

Hi everyone

I am looking for the sequences of the Illumina adapters as well as the primers. I want to make my own library and then send this to a company.
I know the sequences have been published somewhere but I want to make sure I have the right ones.
Thank you for your help.

Odile

[HESmith]

Here's a list of the multiplexing primers:

[odile]

Thanks a lot HESmith. I just found out how much an Illumina oligo kit costs... almost chocked on it. I will order the primers and adapters separately. So, I've been warned about the phosphate on the 1st adapter, but was also told to make sure I had a "Phosphorothioate bond" in the other adapter !!!??? He?....habal espanol? No idea what that is.

[HESmith]

Phosphorothioate is a modified phosphate in which one of the oxygen atoms is replaced by sulfur. The modification makes the phosphodiester bond between nucleotides resistant to exonucleases. Most companies that make oligos can incorporate this modification, which belongs between the 3' and previous nucleotide.

[odile]

I'm learning everyday. Thanks HESmith.

[Marta]

I developed PE indexed adapters for our needs. They worked fine w/o any special modifications, just 5'-P for all PE-1 adapters. We read the index as first 6 letters in each direction. The file can be downloaded from:

http://www.genomecenter.ucdavis.edu/dna_technologies/documents/martas_barcodes.pdf

Marta

[kmcarr]

Marta,

When mixing barcoded libraries in a lane do you find it necessary to make sure that the final pool has somewhat balanced base composition at each position of the barcode? Obviously there is nothing you can do about position seven (T); does that cause an trouble for the image analysis?

[Marta]

I did not see any problems with analysis of unbalanced base composition. The main problem with pooling I experienced is unequal amounts of reads from different libraries in the pool. This is not index specific. Most likely it relates to library quality, errors in concentration estimation, and pipetting small volumes.

[simonandrews]

You need to be careful if you're putting barcodes on the front of standard Illumina reads rather than using their third read barcode method.

The reason is that you end up with very low complexity in the first few bases and this can cause problems with the recognition of clusters by the pipeline.

The way the pipeline works is that it takes the data from the first 4 cycles and uses this to decide where all of the clusters are. If you have nicely separated clusters then you'll be fine whatever sequence you use, but if you use high cluster densities (which you need to if you want to get the most sequence out of a lane), then many of your raw clusters will be composed of two or more overlapping sequences. What the pipeline does therefore is to look for differences in parts of a cluster during the first 4 cycles and uses this to segment mixed clusters into two or more separate clusters.

The problem is that if your sequences are exactly the same for the first 4 bases then you can't separate these clusters so instead of separating them they are either immediately rejected because of their size, or they are kept but later rejected by the purity filter because the sequences will later differ and produce a mixed signal.

We have had libraries where the first 4 bases were the same in all sequences and if we processed these libraries starting at base 5 instead of base 1 we got a huge (>200%) increase in the amount of usable clusters. If you have quite a few different barcodes then the problem will be somewhat mitigated, but if you're using only a small number of barcodes then you might end up losing a lot of potentially recoverable sequence.

What would be nice would be to have an option to do the cluster identification on cycles 5-8, but then go back and call sequence starting at cycle 1, but there is no option to do this in the Illumina pipeline at the moment (and it would require storing 9 cycles of data on the RTA machine, which probably can't cope with that).

[Marta]

simonandrews,

Thank you very much for your posting. I did not experienced/realized this problem yet - the only way I used the indexes is by combining 12 libraries into one pool. In future I may need to have less complex pools and will certainly check for the balance of bases for first 4 nucleotides in the index. Is this all I can do for this type of bar-coding?

You wrote: "What would be nice would be to have an option to do the cluster identification on cycles 5-8, but then go back and call sequence starting at cycle 1"

This is for 4-letter bar-coding w/o A/T ligation? In my case I have 6 letter bar-codes followed by T. Do I understand it right that I would need to identify clusters in cycles 8-11?

[simonandrews]

simonandrews,

Thank you very much for your posting. I did not experienced/realized this problem yet - the only way I used the indexes is by combining 12 libraries into one pool. In future I may need to have less complex pools and will certainly check for the balance of bases for first 4 nucleotides in the index. Is this all I can do for this type of bar-coding?

With the current pipeline I think this is all you can do. The main thing is to be aware that a problem potentially exists and know how to spot it and mitigate its effects. With 12 different sequences you'll be fine since you have an 11/12 chance of being able to distinguish overlapping clusters. Even with 4 bar codes you're OK for 3/4. If you have only 2 or if all your sequences start the same you can start to lose a significant amount of sequence though.

You wrote: "What would be nice would be to have an option to do the cluster identification on cycles 5-8, but then go back and call sequence starting at cycle 1"

This is for 4-letter bar-coding w/o A/T ligation? In my case I have 6 letter bar-codes followed by T. Do I understand it right that I would need to identify clusters in cycles 8-11?

You'd need to start the cluster identification from whichever point your sequences started to diverge. The problem if you leave it too long is that you'll start to get mixed signals even in a clean cluster because of phasing noise - which will get worse the further through your run you go.

The core of the problem is that you need overlapping clusters to have different sequence in order to tell them apart. Anything in your experiment which reduces that diversity will potentially cause you a problem.

[Marta]

Thanks again simonandrews