common reasons for failed barcode reads?

[bombardior]

hi everyone,
i just had my ChIP seq libraries come off a lane from illumina hi-seq (6 samples in there) and was told by my facility that the barcode reads have all failed.
basically i have perfect reads for all 100 bp of my insert, then for some unknown reason when they begin barcode reads nothing happens.
i looked at the raw images and it basically goes from many tiny clear red glowing dots at basepair 100, to almost black blurriness for the 101 base.
(also had low cluster density but they don't think its related)


they're offered to rerun it in a rapid run (to get higher cluster density) and it just came off again with excellent density now but still no barcode reads.

their words "The adaptors are there, but something is interfering with the barcode reads".

"it looks like there might be something wrong with your adaptors (?). We simply didn’t get a barcode read again even though the cluster numbers are up to a reasonable level. Maybe the barcode read primer isn’t priming?"

is anyone aware of any common causes of this? i have friends who also used the same adaptors in their library prep and all have had sucess.

the inserts (without barcode information), i was able to map to my genome of interest and they all look ok (not adapter dimers)

[Brian Bushnell]

Maybe they ran out of reagent, or there was a lighting failure. If the same adapters have worked fine in the past, it sounds like a machine problem.

Edit - oops, didn't notice the same problem occurred twice. In that case, it sounds like the problem IS with your adapters... are they custom? Also, I suggest you look for short insert pairs that read into the adapter sequence, to see if it is what you were expecting.

[bombardior]

the adapters are custom made but according to illumina published sequence. plus it's also been used by other labs and other members of my lab and worked so that's what made it strange....
unfortunately i size selected my inserts to be 200bp so i don't think any reads would go into the adapter sequence..
and yea doesn't sound like a reagent problem also because other people who shared a run with me have theirs perfectly fine.

[Brian Bushnell]

the adapters are custom made but according to illumina published sequence. plus it's also been used by other labs and other members of my lab and worked so that's what made it strange....
unfortunately i size selected my inserts to be 200bp so i don't think any reads would go into the adapter sequence..
and yea doesn't sound like a reagent problem also because other people who shared a run with me have theirs perfectly fine.

Normally, despite size-selection, there is a long tail on each side of the mode. You only need a couple of short-insert reads for this purpose. You can find them quickly using BBMerge like this:

bbmerge.sh in1=r1.fq in2=r2.fq out1=short1.fq out2=short2.fq join=f maxlength=80

Some of those might be false positives, which you could confirm by mapping, but some should be actual short-insert reads with the reverse-complemented adapter in the read.

[GenoMax]

Do you know what the cluster density was for this run? Are your barcodes well separated compared across the same cycle (e.g. no A's in the same position for all barcodes).

Majority of the times when barcodes fail it is because of overloaded samples. The point at which a good run goes over the cliff (as you found out the reads are fine but the barcodes are not) is very narrow.

[bombardior]

Do you know what the cluster density was for this run? Are your barcodes well separated compared across the same cycle (e.g. no A's in the same position for all barcodes).

Majority of the times when barcodes fail it is because of overloaded samples. The point at which a good run goes over the cliff (as you found out the reads are fine but the barcodes are not) is very narrow.

from the first run i had low cluster density of 36M reads, they said they expected about 100M and adjusted accordingly for the second run and i did get 100M the second time.

[GenoMax]

The cluster density number is expressed as NNNN k/mm^2. This refers to how many clusters there were per mm^2 on the flowcell, which indicates loading density.

You are not going to see/know this number in your data. You will have to ask the facility (sounds like you got the sequencing done some place else). Have they given you this data or have they just told you that the barcodes have failed?

Are you not using standard illumina barcodes?

[bombardior]

i think i may have used the wrong PCR amplification primers....
i used these following:
Illumina Single End PCR Primer 1 AATGATACGGCGACCACCGAGATCTACACTCTTTCCCTACACGACGCTCTTCCGATCT
Illumina Single End PCR Primer 2 CAAGCAGAAGACGGCATACGAGCTCTTCCGATCT

it looks like this is old technology before multiplexing is allowed? i was the first person to use a new batch and the person may have given me the old sequence to order.

[bombardior]

what i noticed from the above two sequences is that they both have the same 3' end.... meaning they're not amplifying from different ends?

[bombardior]

OMG...... i think i've figured it out................
what was sent to me, and labeled as PCR primer 1, is actually the universal adaptor...................... i'm going to kill my lab mate!!

[SNPsaurus]

Yikes! So did you get weak amplification of primer 2 binding at the end to the primer 1 binding sequence? Hmm, I'm not sure how you'd get clusters formed from this!

[bombardior]

this is really strange... so the primer 1 used in my PCR amplification is actual the universal adaptor, and primer 2 has the first sequence as ACTUAL primer 2 but the 3' end is also the same as the 3' end of the universal adaptor..... so what ended up happening is everything is amplified except the index/barcoded region and i had regular sequencing runs... except there was no place for the barcode primer to anneal and read...

thanks everyone for the helpful suggestions though! i'm glad it's figured out at least...