Problems cDNA sequencng

[Alexander Tchourbanov]

Dear colleagues,

We are facing the problem with cDNA sequencing. Our read length is significantly shorter than the expected average size. Interestingly enough two other quarter plates in the last cDNA run contained DNA samples and went perfectly fine. At this point we suspect the problem is the cDNA synthesis, however, the last run cDNA has been obtained from a very reputable place and all the quality assessments were fine, including fragment length distribution after the rapid library preparation. The beads recovery rate was 7%. Can someone point to the possible omissions that we might have or recommend some reliable ways of cDNA synthesis compatible with 454 GS-FLX runs.

Thanks!

Alexander

[Guedes]

Alexander,

I'm going to work with 454 FLX too. I'll use the " Smarter cDNA synthesis" kit to do my cDNA. What kit did you use to synthesize cDNA? After synthesis, have you analised it using a Bioanalyser? About reads lenght, what was the average you obtained? I'm very interested in these informations because I don't know much about this.

Thanks

Fernanda

[Alexander Tchourbanov]

cDNA synthesis

We used Evrogen. Something did not work quite well, we try to figure out what could be a problem. Bioanalyser reported no problems, all length distribution was in a Roche recommended interval.

[pmiguel]

I think poly A tails are the source of nearly all woes with 454 cDNA sequencing. I have posted about this previously:

http://seqanswers.com/forums/showpost.php?p=11587&postcount=10

The problem is that those "V" anchored oligo dT primers can still prime in the midst of the poly A tail. It might sound unlikely, but when we were trying to get 454 cDNA libraries to work well, we Sanger sequenced sets of clones from several libraries created for the 454. Despite using the now nearly canonical "V" anchored oligo dT primer, it was obvious that many of the libraries nevertheless contained long poly A and poly T tracts. The amount of sequence obtained when these libraries were run on the 454 was roughly inversely proportional to the number of "long poly A" clones we saw via Sanger sequencing.

If you use a cDNA/amplicon creation technique that forces the poly A away from the A adaptor side (like the one developed by Metz, Mockaitis and Buchannan-Carter et al.) results are much more robust. Sometimes better than one gets with genomic DNA libraries.

The other issue here is one of Roche shooting themselves in the foot by failing to provide a metric that would tell a 454 user that they poisoned their run with long homopolymer tracts. It might be difficult to implement such a metric but their failure to do so wastes resources downstream. I'm sure that nearly every other support call they get might be the result of this phenomenon.

Currently the only way I know to see this phenomenon for sure is to actually fire up gsRunbrowser and iterate through the early bases on the full PTP image viewer -- the ones after the 4 key bases (and MIDs, if any). If you are seeing large percentages of your wells flaring on the "T" flows, you know your sequence yield is not what it could be.

--
Phillip

[martin2]

Even such reads we have from cDNA protocol using Zn2+ and high temperature for cleaving of the sample (polyA-tailed mRNA). Yes, after the sequencing run there was almost no dT left in the falcon tube, unlike dG and dC, especially. The avg read length was about 230 only, and have heard from other they had also bad results.

>blah
TTTTTTTTNTTTTTTTTTTTNTTTTTTTTTTTATTTTTTTTTTTATTTTTTTTTTTATTT
TTTTTTTTATTTTTTTTTTTATTTTTTTTTTTATTTTTTTTTTTATTTTTTTTTTTATTT
TTTTTTTATTTTTTTTTTATTTTTTTTTATTTTTTTTATTTTTTTATTTTTTATTTTTAT
TTTATTTTATTTATTTATTATTATTATATATATATATATATATATATATATATATATATA
TA

I suspect such well ate up all the nucleotides available in a flow.

[dagarfield]

Just one quick suggestion: Check your sequences for the presence of the cDNA primers. I've processed several libraries generated using the Smart kit and ended up with suspiciously short sequences that were filled, essentially, with Ts, As, and Smart kit primer sequences.

I imagine similar kits may sometimes generate similar artifacts.

[pmiguel]

Just one quick suggestion: Check your sequences for the presence of the cDNA primers. I've processed several libraries generated using the Smart kit and ended up with suspiciously short sequences that were filled, essentially, with Ts, As, and Smart kit primer sequences.

I imagine similar kits may sometimes generate similar artifacts.

About the adaptor concatamers sometimes seen in SMART kits, as I have think I know what cause them: (link)

--
Phillip