View Single Post
Old 04-17-2009, 04:43 AM   #7
Senior Member
Location: USA, Midwest

Join Date: May 2008
Posts: 1,165

Originally Posted by gerald2545 View Post
we run only few project on our 454 (titanium only) and we were surprised about the amount of filtered reads and the length of the high quality reads. could you please tell me (if possible) the mean length and the percentage of passed filter?

For example, for our last run (4 regions, cDNA from chicken and duck), for 2 regions (chicken) we only have ~15% of passed filters (whereas Roche expects more than 50% for a titanium run - information from GSFLX Titanium metrics, but it may be for a genomic DNA project). Problems came from dots (cf attached spreadsheet).


Your filtering output does not surprise me at all. We have seen very similar results with many of our cDNA runs on the 454. In the past, with the GS FLX Standard, most failed reads were rejected by the Mixed filter; now with the Titanium it seems that more are rejected by either the Dot or Short Quality filter. I don't know if this a difference in the sequence itself or in the filtering algorithms.

Our read lengths for cDNA samples often show a bimodal distribution, with one peak at the expected 450-500nt range and second peak in the 100-150nt range. This results in a lower mean and median read length but those statistics don't give the full picture in these cases.

Remember that all of the performance numbers which Roche publishes (e.g. 50% filter passed reads, 400-450nt mean read length) are based on sequencing very well behaved genomic DNA.

Our protocol for preparing cDNA for 454 (SMART kit with modified first strand primer) was originally developed when our instrument was a GS-20, then carried through to the GS FLX. It may be a time to develop a new strategy for the Titanium.
kmcarr is offline   Reply With Quote