Los Angeles, CA February 16th 2014 – Last Sunday, we posted a blog on the highlights of The 2014 Advances in Genome Biology and Technology (AGBT) conference, as perceived by attendees. We received a number of interesting comments (most notably from Kyle, a NGS data analyst at a top 5 national sequencing center) on this initial blog entry; as a result, and at the request of a few readers, these comments are presented largely unedited below:Genomic Platforms

• Illumina will likely retain dominance in the high output sequencing market. However, this throughput may be more than many customers in the targeted diagnostics, clinical, and small research labs may need. While Illumina has made great progress in improving the speed, efficiency, and output of their instruments, and demonstrated a remarkable ability to scale up, it is yet to be seen if they are equally efficient at scaling down. Another interesting undercurrent with Illumina is that the company is offering fee for service sequencing at much less cost than many sequencing centers and lab cores charge. This may have the effect of cannibalizing their reagent and instrument sales and even puts larger sequencing operations (Illumina's biggest clients) at risk.
• Ion Torrent continues to work through problems with their technology driven by errors near homopolymer sites. The P II chips continues to face headwinds, and we don’t hear about the P III chip anymore. The Ampliseq line is expanding to include several targeted diagnostic panels for a variety of diseases interrogating DNA for both mutations and copy number, and RNA for differential expression and gene fusion detection. Several other technologies for targeted gene capture exist and were discussed at AGBT, but Ampliseq has the advantage of working with very low amounts of DNA (or RNA). While DNA amounts are generally not limiting in many clinical setting, they are incredibly limiting for retrospective analyses. Thus, Ampliseq is poised to offer products which could be tested and gain regulatory approval via large, retrospective datasets.
• Several presentations focused on the uses of PacBio ultra-long single-molecule read technology. No presentation was remiss in mentioning the high error rates (primarily insertions and deletions) seen with this technology. Nevertheless, PacBio data was shown applied to many applications such as genome assembly, RNA-seq, or cancer biology. While interesting and important findings were presented concerning RNA-seq and cancer biology the principal apparent use for PacBio data is to aid in draft genome scaffolding and assembly. PacBio may soon become the standard tool for genome assembly, with short read technologies being used to correct for higher error rates and SNP calling. The technology remains far too expensive to apply to high throughput applications in a robust and high throughput way.
• NanoPores (such as ones offered by Oxford Nanopore) have the advantage of being extremely inexpensive to start generating data with. For the cost of the chip (advertised at $1,000) one can begin to generate ultra-long read data, though the yield was not entirely clear. This is a substantial cost reduction, in comparison to the ~$1M capital equipment cost required to acquire a PacBio instrument.
• With regards to Genapsys’ Genius system, a ~$2,500 sequencing device could be a game changer and open the NGS market for many research labs, diagnostic and environmental testing groups, hospitals, and even the general public (though what they would do with a sequencer is an interesting question). However, so little is known about the device or the technology that everything is just speculation at this point.
• BioNanoGenomics demonstrated the Iris which can be used for genome finishing. BNG does not actually generate any sequence data but rather maps of restriction or nickase sites across the genome. While this technology could be used for genome assembly and perhaps have some role in studying or identifying structural rearrangements in cancer, prenatal diagnostics, or other diseases, other platforms will likely dominate these markets.
• While Quantum Biosystems did not give a presentation or presented a poster, they did show the evolution of their chip at the bar, and openly discussed the technical details of their platform

Library preparation

• On the library preparation end, GenCell and its CLIC platform may generate a significant interest from labs that require high throughput (see our article on this platform)
• Illumina’s NeoPrep was highlighted as an interesting sample preparation technology
• Caliper (PerkinElmer) appeared to offer a flexible instrument, but priced at a premium
• Many large companies are offering targeted panels such as Ampliseq, NimbleGen, Agilent, Qiagen, IDT, and Enzymatics. At the moment, Ampliseq dominates for (low) amount of input material required to generate a library and average cost per sample (should all of the reagent be consumed) while NimbleGen can generate larger panels and at lower cost per sample for smaller studies. Ampliseq has married their technology to the Ion platform both in reagent design and marketing, making Ampliseq largely inaccessible to those who do not own Ion sequencers

Clinical Sequencing

• Several groups presented on the use of NGS technologies for clinical diagnostics. These included diagnostics using targeted, actionable gene panels for Mendelian disease and cancer, prenatal diagnostics, and even environmental monitoring for drug resistant infectious diseases.
• As a market, clinical diagnostics using NGS technology appears highly fragmented. Many clinical labs are offering competing tests, yet failing to offer an entire suite of tests for all purposes. In my opinion, there are great opportunities for consolidations and mergers in this field in the near future.
• Several approaches to prenatal diagnostics were discussed both for detection of disease in the fetus and for pre-implantation screening. Perhaps the most exciting was for the detection of trisomy and copy number variation detection from maternal plasma samples by Sequenom
• Ethical considerations of clinical NGS data were not really addressed in this meeting

Bioinformatics

• Though notable by its absence, data management issues were present as footnotes in many presentations at AGBT. Many institutes, diagnostic labs, and companies are facing data management and analysis problems due to the amounts of data generated by NGS technologies. Many groups have moved to the "cloud" and are using the Amazon EC2 services for data storage and computing while others continue to work on ever growing file systems handling several petabytes of data. Management, tracking, and storing this data for large analysis projects is becoming more challenging than developing variant calling algorithms or designing experiments. While no clear solution to this problem was advanced, there is clear consumer need for solutions to the data deluge
• On the bioinformatics end, Jason Chin and Eugene Myers gave outstanding presentation on PacBio assembly

Ultimately, while we have issues with sample prep, the real NGS bottleneck is with data analysis. We can generate data much faster than our customers can analyze them---

Authors: Stephane Budel, Partner at DeciBio, LLCConnect with Stephane Budel on Google+https://plus.google.com/+StephaneBudel