We've actually gone as high as 30 cycles. Typically if samples are below quantitation range we try to move only half of the library forward to the PCR step and do 15-20 cycles (depending on your mood that day). Based on those results we may repeat the PCR with the remaining sample and increase the cycles by some arbitrary amount. We do it that was to avoid reamplifying the amplified material but we've done that too when we have to. This isn't something that I'd say is routine. But it's come up a number of times when we've been virus fishing in clinical samples. After depleting the host as much as possible you are usually going on blind faith that something is there. The other curious thing that we've encountered when doing excessive cycles is short insert contamination. We don't see this when doing lower cycles and I'm assuming it is because the smaller fragments amplify more efficiently. These are not always apparent on the trace but there is a distinct population of reads between 50-250 bp (mode ~ 150 bp) in the final data sets when the main library is typically 500+ bp. In fact we Pippin the libraries with a gate range of approx. 500-800 bp so why the hell there are still small fragments is beyond me. We're still trying to arrive at a reliable and consistent way of removing this material. It's not that it isn't good data but it often causes runs to fail due to over clustering. The flow cells are loaded based on molar calculations and if you are not accounting for the small stuff you can be loading 5 or 10 times more than you think. I know this discussion has been somewhat outside of the scope of this forum but then again understanding the wet lab side of things can give you a better understanding on the nature of the data. Shaun From: Andrej Benjak <abenjak@xxxxxxxxx> To: mira_talk@xxxxxxxxxxxxx Date: 2014-11-04 01:59 AM Subject: [mira_talk] Re: a question about Nextera Kit vs TruSeq Kit Sent by: mira_talk-bounce@xxxxxxxxxxxxx We are often dealing with small starting material and found the TruSeq Chip library kit is much better than the standard TruSeq. It can handle 5-10 ng of DNA, but we have done it with less (even from what appeared to be empty tubes!). We normally do 18 PCR cycles. I have never checked the read coverage distribution in detail since we mostly use our sequences in mapping analyses, but from looking at the genome browser it looks ok. What can happen with very small starting material is the higher occurrence of duplicate reads (PCR duplicates), which has to be considered when assessing the sequencing coverage, but perhaps this is our special case because most of our DNA derives from the host, and the bug we are interested in is poorly represented. Best, Andrej On 11/04/2014 02:31 AM, Kevin Chen wrote: Hi Shaun, thank you very much for your explanation. I talked to our lab manager and we are very interested to test out the minimal amount that TruSeq library can handle. Do you have any suggestion on cycle number and the starting amount of material recommendation? For example, what number of PCR cycles works for an amount of like 100ng of DNA? I understand it is empirical but it would be huge help for us to start. Another question is when you measure the depth of coverage, what metrics you used to measure the skew in depth of coverage? I am thinking of coefficient of variation, but it might be over simplifying the measurement. Thank you very much! On Wed, Oct 29, 2014 at 2:02 PM, Shaun Tyler < Shaun.Tyler@xxxxxxxxxxxxxxx> wrote: We've successfully made TruSeq libraries when amounts have been so low we've been unable to quantitate. It just requires playing with the PCR cycles and going on faith. So the minimum requirement isn't really the "minimum". Something to keep in mind with Nextera is that it may only require a small amount of input but you will still need to obtain larger amounts from your samples for good results. The yields need to be high enough to accurately quatitate. Because the tagmentation reaction is very much ratio dependant if you stray from the recommended input amounts you can shear it way too much or not enough. Either way the sequencing results will be poor. This isn't as much of an issue with TruSeq as the sonicating isn't affected by concentration that much. We've also found with the Qbit that concentrations below about 1 ng/ul (using the low range standards) really can't be trusted. So if you are anticipating very low yields from your samples TruSeq might still be the best way to go for consistent results. We've done hundreds of both types of libraries primarily on bacteria and there is a noticeable skew with Nextera. If you map the reads back to the assembly (or are using an assembler that produces more than just a fasta file) it is quite apparent. 2 to 3 fold differences in depth of coverage are not uncommon (Bastien originally helped sort this out). Because of this we've been primarily using SPAdes as it does not seem to be adversely affected by the uneven distribution. However what we lack in the resulting assemblies is any reliable information regarding copy number for repetitive regions. The depth of coverage gives you some indication but because of the skewed distribution even that can't be completely trusted. Other than the input requirements between the two approaches the biggest attraction to Nextera XT was the cost. It was significantly cheaper per sample than TruSeq and when your planning on a couple of hundred samples that was a significant concern. But Illumina recently slashed their prices on TruSeq so that's not as big of a factor any more. They are more labour intensive and involve more in consumable costs but we're starting to migrate more towards TruSeq now because of the cleaner data. BTW - cDNA was mentioned which implies transcriptome analysis in which case I would definitely NOT use Nextera XT. With the distribution skew you would never be able to trust anything other than genes that are substantially up or down regulated and any quantitative measurements would just be ball park estimates. At least that's my opinion ;-) Shaun Inactive hide details for Kevin Chen ---2014-10-29 11:17:41 AM---Nextera XT uses 1ng gDNA or cDNA. That is a huge advantage. AdKevin Chen ---2014-10-29 11:17:41 AM---Nextera XT uses 1ng gDNA or cDNA. That is a huge advantage. Adrian, can you give more details on the From: Kevin Chen <wchen20@xxxxxxxxx> To: mira_talk@xxxxxxxxxxxxx Date: 2014-10-29 11:17 AM Subject: [mira_talk] Re: a question about Nextera Kit vs TruSeq Kit Sent by: mira_talk-bounce@xxxxxxxxxxxxx Nextera XT uses 1ng gDNA or cDNA. That is a huge advantage. Adrian, can you give more details on the coverage consistency, what metrics you used to measure it? Thanks! Regards, Bing On Wed, Oct 29, 2014 at 10:41 AM, Chris Hoefler <hoeflerb@xxxxxxxxx > wrote: How minute are we talking about? TruSeq Nano can start with ~100 ng unsheared gDNA. On Tue, Oct 28, 2014 at 7:03 PM, Kevin Chen < wchen20@xxxxxxxxx> wrote: Hi, Bastian & Mira community I am Bing from U of Maryland. When I read your Mira manual, I notice this statement below. I am interested in getting more information on that. Because Nextera kit has its own advantages that TruSeq does not have, for example, working with minute amount of material, I have high hope for Nextera. Do you have any prelim data to support this statement? If this is true, I have to re-think a lot of my experiment design using Nextera, that’ll change a lot of things. "For de-novo assemblies, do NOT (never ever at all and under no circumstances) use the Nextera kit, take TruSeq. The non-random fragmentation behaviour of Nextera leads to all sorts of problems for assemblers (not only MIRA) which try to use kmer frequencies as a criterion for repetitiveness of a given sequence." Thank you in advance, Bing -- Chris Hoefler, PhD Postdoctoral Research Associate Straight Lab Texas A&M University 2128 TAMU College Station, TX 77843-2128