Repetitive reads

[obSEQuious]

Hi,

I just got data back from a Solexa run of a ChIP for a histone modification occuring in euchromatic regions of the genome. My tags, though, look like this:

AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA 200126
CACACACACACACACACACACACACACACACACACA 9848
CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC 3245
TGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTG 3211
ACACACACACACACACACACACACACACACACACAC 2684
GAGAGAGAGAGAGAGAGAGAGAGAGAGAGAGAGAGA 1390
AGAGAGAGAGAGAGAGAGAGAGAGAGAGAGAGAGAG 1226
GGTTAGGGTTAGGGTTAGGGTTAGGGTTAGGGTTAG 1105
TGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTT 905
CACACACACACACACACACACACACACACACACACC 591
GATCGGAAGAGCTCGTATGCCGTCTTCTGCTTGAAA 582
CCTAACCCTAACCCTAACCCTAACCCTAACCCTAAC 513
GGGTTAGGGTTAGGGTTAGGGTTAGGGTTAGGGTTA 446
TGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTTTT 406
CTAACCCTAACCCTAACCCTAACCCTAACCCTAACC 386
CACACACACACACACACACACACACACACACACACT 373
CACACACACACACACACACACACACACACACACAGA 322
GGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTG 312
CCCTAACCCTAACCCTAACCCTAACCCTAACCCTAA 299
TGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTGTTTG 294
GACAGACAGACAGACAGACAGACAGACAGACAGACA 290
CACACACACACACACACACACACACACACACACAAA 290
GGGGCAGAAGCTGCCTGAAAGGTGCTTGAGCAACGT 285
TACACACACACACACACACACACACACACACACACA 268

And go on like that. The majority of my reads are either are or are almost straight runs of a single base or or dinucleotides. Only 5.5% of my reads mapped to the genome at all, and those that did are not where I biologically expect them to be. Blatting a lot of them shows that they are usually in repetitive elements.

Chromatin was fragmented by digestion to mononucleosomes, but the bioanalyzer trace shows the existence a of large peak at ~300 bp, which is larger than the expected peak at ~180.

So I'm trying to figure out what when wrong with the run, so this can be avoided in the future. It looks like something's contaminating my sample, but blasting these reads doesn't show any obvious answers. Thoughts?

[Torst]

It appears there is one or more problems with the sequencing quality. I believe with Solexa that "A"s will be called when the flow cell is over illuminated or too oily and so on (User ScottC on this forum can probably explain this better!). That is why you have a large proportion of poly-A reads. You probably expect these in some proportion, but too many of them suggest a quality control issue.

The repetitive dimers look very suspicious too. I don't have any explanations for those, though.

[bioinfosm]

We saw similar behavior from a recent ChipSEQ. Less than 10% aligned to the actual reference, while 1% or so aligned to human dna, 1% or so formed contigs that align to some bacteria, etc., 1% aligned to adapter/primer sequence, but no clue of the leftovers!

Using 25bp reads for ChipSEQ data sounded more reasonable in our case, but still a huge amount of reads are not accounted for yet..

[elisa*_*]

poly A artefact near the edge of a tile

I came across this info in the release note of Maq 0.6.4.: "It is important to note that Illumina/Solexa sequencing may produce many false polyA at the edges of a tile. These polyA artefacts may greatly increase the running time of maq. Users are advised to remove these artefacts with their own scripts before alignment. For the moment maq does not provide a general functionality for filtering polyA."

Does anyone know about the source of this artefact?

[jlli]

Does SOLiD sequencing generate these artificial repetitive reads near the edge of a slide? We observed similar behavior from our SOLiD ChIPSeq project. We have no clue where went wrong.

[elisa*_*]

Hi jlli, I don't know the reason for Solid, but I did find an explanation for Solexa given by SillyPoint in the following thread. Maybe you have similar problems.

http://seqanswers.com/forums/showthread.php?t=358

[acnoll]

Assuming you had a control lane on the FC how did it look - if you saw the same polyA problem this would argue for the oil theory. Was there one or more libraries constructed for the original sample? If the same result appeared in more than one library this would eliminate amplification bias.