454 DNA Sequencing
Multiple Samples, Run Expectations & Guarantees
Physically dividing a plate:A Titanium sequencing plate is always split into at least 2 regions.It is possible to physically split the plate into as many as 16 regions, but splitting the plate into 16 regions results in obtaining half as many sequences (overall) as a full run with only 2 regions.Splitting a plate into 4 regions still gives many reads, but splitting more than that significantly reduces total output.Thus, for researchers that want less data than can be obtained on a quarter plate, we recommend using MID-tags to identify individual samples and to pool multiple samples into one region.We will combine samples among multiple researchers to allow everyone to achieve the number of reads and lowest cost possible.
MID-tags:Roche has recently released primer sequences for 151 MID-tagged linkers.Roche does not yet (as of August 2009) have linkers available, individual oligos must be purchased from manufacturers (e.g., IDT, Sigma, Invitrogen, etc.).We have 10 Roche-specified MID-tagged linkers available.We have modified the library prep protocols to make use of unmodified oligos and now have many more MID-tag linkers made from unmodified oligos.We also have a set of 12 “SimpleX” linkers available.Used combinatorially, this gives us up to 151 x 12 = 1812 indexes available for immediate use.The number of SimpleX tags possible is very large, so really we have more or less a limitless number, much more than would ever be needed when using the SudokuDNA strategy.Use of MID-tags and especially combinatorial tags, will however, still result in a slight loss of reads relative to a single sample.
We are contractually obligated (by use of Roche reagents for sequencing) to conduct at least 10,000 sequencing reactions.We interpret that to mean we can split a full run into no more than ~100 MID-tagged samples.If users provide us with pooled samples (e.g., already sub-tagged with SimpleX linkers) for each sample we will MID-tag, then fewer reads/sample can be achieved.The smallest project that we can currently support is approximately equivalent to 1/16th of full-run, but we will achieve this using MID-tags and pooling.
cDNA: Roche does not support cDNA sequencing with Titanium chemistry, but several sequencing centers have been working on this problem.We have adapted the protocol in use at NCState and achieved good results.Researchers should expect a slight reduction in both the number of reads and read-length relative to shotgun reads.Read length is impacted by short templates (they are real and you want them).The number of reads is reduced slightly (~10%) due to runs of poly-A and short-templates that don’t sequence well and ruin the reads in adjacent wells.Because Titanium chemistry is not well suited to short reads (i.e., <200 bases), researchers who want to investigate microRNAs or other short RNA transcripts will likely want to use a shorter read length platform (i.e., Illumina Genome Analyzer or Applied BiosystemsSOLiD).
Roche provides guidance in the number of reads (sequences) researchers should expect from various configurations of the 454 using Titanium chemistry, and assuming a single shotgun library prep from genomic DNA.As explained above, MID-tags, cDNA, and other factors will reduce the number of reads obtained.
Table 1. Minimum expectations users of the GGF should have for the number of reads (sequences) obtained from the 454:
| Type of Library |
Number of reads
|
|||
|---|---|---|---|---|
| Full run | Half-run | Quarter-run | Minimum | |
| Shotgun Library, single sample | 900,000 | 450,000 | 180,000 | 10,000 |
| cDNA, single sample | 800,000 | 400,000 | 150,000 | 10,000 |
| MID-tagged, 8 samples (total reads) | 800,000 | 400,000 | 150,000 | 10,000 |
| cDNA, 8 MID-tagged samples | 700,000 | 350,000 | 125,000 | 10,000 |
If a run fails totally (less than half the expected number of reads), then the libraries will be remade & another run done at no expense to the user. If at least half the expected number of reads are obtained, then researchers will be billed/invoiced proportionally (e.g., if 800,000 reads are obtained from a single shotgun library full run, then 800000/900000 = 0.888 x run cost will be invoiced).
