HiSeq 2000 v3: Bad tiles?

[rkirkbride]

Our lab is fairly new to the NGS game, but we recently sent off some mRNA seq samples which were prepared with TruSeq barcodes.

We obtained the data in the form of 48 qseq files with 100 bp reads, and a matched set of 48 qseq files containing the 7 bp barcode sequences.

About half of the files look fine, while the other half have barcodes that are unable to call about half the 7bp sequence, and low Phred Scores both in the 100bp sequences and in the barcode sequences.

It is my understanding that each of these files correspond to "pseudotiles", and I was wondering what that means in practical terms of position within the flowcell lane.

I know which are good and which are bad, and was wondering the following:

1) If anyone knows how file numbering and tile position are related.
2) What might cause some "pseudotiles" to have problems while others appear fine?

Thanks!

-Ryan

[SillyPoint]

I can do the easy one:

It sounds like you have one lane's worth of data. A HiSeq lane is processed as 48 tiles.

I haven't heard the term 'pseudotiles', but I'm guessing that just refers to the fact that the HiSeq images continuously, rather than taking discrete snapshots as the GAII does, and so the division of image data into batches is an arbitrary artefact of the image-processing software.

Presumably your tiles have 4-digit numbers of the form [12][1-3]0[1-8]. The first digit is flowcell surface: 1=bottom, 2=top (I think, or vv). Second digit is 'swath': a row of 8 tiles extending the length of the lane. A swath is actually imaged in one continuous pass. A lane surface has 3 adjacent swaths. Digit 3 is 0. Digit 4 is tile# within the swath.

Your other question is far harder, of course. Keep in mind that each tile is separately post-processed, starting with images and applying registration, crosstalk and phasing corrections, then computing quality scores. You're on the right track in wanting to look for spatial correlation in the tiles qualities.

HTH. Good luck.

--TS

[rkirkbride]

Thanks, that was quite helpful, I was having a heck of a time finding details about how the imaging was carried out. Based on this info, it looks like one end of the flow cell is responsible for the problematic files, which helps narrow things a bit.

Cheers!

-Ryan

[simonandrews]

Could it be related to this?

http://seqanswers.com/forums/showthread.php?t=13830