high first base error rate

[bioinfosm]

Hi,

I have noticed this with all of the recent RNA seq runs we have had with the solexa. The first base error rate is very high (looking at error graphs). Attached is the image for one of the tiles..

do people face the same issue? Any remedies, apart from skipping the first read for eland alignments I suppose!

I think we see this with other solexa runs as well, not just rna-seq..

thanks

[BaCh]

do people face the same issue?
I think we see this with other solexa runs as well, not just rna-seq..

I've just checked a couple of genome sequencing runs from 2008 (36mers) and 2009 (72mers). I have to clip bases on the left of ~4.5% to 6% of the reads (can't tell you how much, the statistics is not that detailed in the log file), the rest has no errors in the first ~20 bases.

Any remedies, apart from skipping the first read for eland alignments I suppose!

Just clip away what's really wrong. I (well, MIRA) make an all vs. all comparison of the reads and clip away read ends that are seen only once or which are not confirmed by sequence from a reverse read. Takes just a couple of minutes for one lane.

On the other hand ... with enough data, always throwing away the very first base probably does not hurt.

Regards,
B.

[NSTbioinformatics]

The first base is artifact "T" from the illumina protocal. The updated protocal does not have the problem.

You should use "USE_BASES nY*" in the Gerald configure file to ingore it.

[bioinfosm]

The first base is artifact "T" from the illumina protocal. The updated protocal does not have the problem.

You should use "USE_BASES nY*" in the Gerald configure file to ingore it.

Could you elaborate on what is artifact "T", and which updated protocol you refer to that overcomes this issue?

Thanks

[NSTbioinformatics]

In our illumina runs, the first base "T" occured only on phix lane, not in other lanes (our owe sequeneces). As i heard from our technician, the first base is artifact "T" from the illumina protocal.Sorry i don't know much detail about it. Now we do not have the first base T in the phix lane. Both the illumina protocal and the phix reference are updated.