No enrichment option

[Michael L. Altshuler]

Hi Colleagues,

If I am not behind the times, none of the currently available sequencing platforms can manage the entire human genome as single sample in single lane.

At the same time, there is an option of running aliquots from the same library in several lanes. The data collected from individual lanes is merged for downstream analysis or comparison.

Running aliquots from the same library in the eight available lanes of one HiSeq or Genome Analyzer IIx flowcell will enhance the coverage of the entire human genome up to 30-fold (see the Illumina spec sheet for HiSeq 2000).
This coverage might be enough to bring forth the accurate and reproducible data on any significant part of the human genome.

Thus, the multilane approach might enable skipping the enrichment entirely or simplifying it.

I realize that it would add considerably to the price/running cost of the analysis. However, the enhanced price of taking up the several lanes for single library might be balanced by reduced occurrence of failed runs and, also, by taking a good deal of money and trouble off the enrichment.

I am interested in personal experience as well as in a reference to any publicly available information.

[krobison]

The claim on the SOLiD 5500xl (nee SOLiD 4hq) is a single human genome on a single flowcell. Illumina makes the same claim with HiSeq. I don't believe any HiSeq genomes have been published yet, though there must be a large number in process.

On Illumina (SOLiD doesn't have lanes), what you describe is what I understand is standard operating procedure.

The point of enriching for a targeted subset is to pack more samples per run at a modest increase in running cost (the enrichment) at the price of losing some data. Whether this tradeoff is worthwhile is something that must be decided on a project-by-project basis.

[Michael L. Altshuler]

Many thanks Krobison,

Perhaps I should have been more explicit in formulating my question, and mentioned Genome Analyzer IIx instead of HiSeq. Below is an attempt to spell it out:

I am concerned about the complexity of sample preparation for GAIIx and its vulnerability to errors. My idea is that recruiting several lanes instead of one will make sample preparation more robust.

In other words, the increase in coverage resulting from running aliquots of the same library in multiple lanes could make up for some errors in sample preparation.

Moreover, this approach could be as beneficial as to enable skipping some of the steps in the standard sample preparation for GAIIx.

I hope this time it is comprehensible.

Another consideration in favor or against the use of the multiple lanes is the expense. The cost of using extra lanes could possibly be compensated by the savings on failed runs and the simplified process of the library preparation.

Of course, it is not a new idea [e.g., http://seqanswers.com/forums/showthread.php?t=5468 , http://biostar.stackexchange.com/que...out-bam-files; “Molecular Diagnosis of Neonatal Diabetes Mellitus Using Next-Generation Sequencing of the Whole Exome”: http://www.ncbi.nlm.nih.gov/pmc/arti...6/?tool=pubmed ]. I am sure it has been implemented many times. It is usually mentioned in a casual and matter-of-fact way. I find it difficult to find anything substantial about it.