Blog @ Illumina
Real scientists. Real commentary.

Changing Landscapes of Medical Genetics and Microbiomes- American Scientific Summit Day 2

Amy Cullinan Ph.D
| Jun 17, 2013

describe the imageThe second day of the American Scientific Summit kicked off a bit early, especially for those of us who stayed up late to enjoy the excellent food and entertainment on Bourbon Street. The two themes of the morning were the changing landscape of medical genetics, and microbial detection and epigenetics.

First up was Richard Gibbs from the Baylor College of Medicine Human Genome Sequencing Center on the realities of clinical sequencing. He started off talking about predictability of future events and their impact on a fast-moving field like genomics, with the central question of what will be the drivers for progress. Dr. Gibbs presented integrative genomics examples such the “Snyderome” and modeling common/complex diseases, highlighted the importance of rapid  variation tests and getting to actionable findings for cancer genomics (covered extensively in my previous post), and added a few more gems to the wealth of evidence that literally everything we consume, metabolize, and excrete affects and is affected by microbiomes. Moving on to  drivers for Mendelian disease, Gibbs talked about a new clinical exome sequencing test launched at Baylor in 2011, in which over 1,700 cases have been sequenced to date using an integrated analysis pipeline. For results based on phenotype, the whole exome results report calls deleterious mutations, variants of unknown clinical significance (in genes associated with the clinical phenotype), and medically actionable mutations. Data from exomes studied to date show that one quarter of disease conditions show de novo and novel variants, and that they are able to dissect situations in which a combination of alleles explain the phenotype. For example, four subjects with independent de novo truncating mutations in the same ASXL3 gene define a new disease phenotype that overlaps with Bohring-Opitz syndrome. Dr. Gibbs emphasized the importance of medical records over time as technology accelerates, and that we must find ways to standardize reporting and link these through with the proper information to make all this genome information useful.

Liz Worthey from the Medical College of Wisconsin spoke next about the progress and challenges of making a definitive diagnosis with WGS and the path to next-day clinical delivery. She introduced a pilot study of WGS for clinical diagnosis in a pediatric setting, in which 3-6 cases were reviewed monthly, and over the entire 18-month study, 23 had been sequenced. They achieved a genetic diagnosis in 26% of these cases, and 4 additional cases showed variants of unknown significance. Dr. Worthey pointed out that in some cases, the actionable information is to not undertake a particular treatment, citing the case of a newborn presenting with excess ammonia production, and was started on treatment that seemed to improve the condition. However at 10 weeks, the infant showed acute liver and neuronal complications. Sequencing pinpointed two pathogenic variants in a gene called TWINKLE that caused mitochondrial DNA depletion syndrome with secondary liver failure. The decision to avoid a costly and ultimately futile liver transplant was made, and although the child sadly did not survive, it opened up an avenue for family testing in the future. Dr. Worthey described a robust pipeline that provides whole-genome sequencing from sample drop-off on the morning of day 1 to completed report on the evening of day 2. Remaining challenges include overcoming technical hurdles with sequencing gaps and changing reference genomes, the mechanics of data storage and retrieval, and inconclusive phenotype-to-genotype associations. Despite all of these challenges, Dr. Worthey remains optimistic that NGS-based genomic medicine provides keen insight to human health and disease, identifying genes and pathways and making clinical diagnosis today a reality.

Ken Bloom, chief medical officer of Clarient Inc., spoke on the challenges of cancer pathology with an aptly-titled talk “Think Big, Start Small”. Reviewing the job description for today’s pathologist, Dr. Bloom reminded us what diagnosis means for the patient and the efforts they and others strive towards for changing the course of these diagnoses. Dr. Bloom talked about data from IPASS, the Iressa (gefitinib) Pan-Asian Study for non-small cell lung cancer, showing that treatment increased survival rates of patients with certain EGFR mutations (exon 19 del, L858R), but had the opposite on those with other mutations (T790M, exon 20 ins) in the same gene. A large number of other mutations exist in EGFR, and the treatment response here is unclear. Dr. Bloom mentioned that assessing genes with panels, exomes, or whole-genome strategies provides more comprehensive information when compared to single gene or hotspot tests, but increased cost and complexity also exists. Using the example of BRAF codon 600 mutations in melanoma, provides evidence that NGS approaches can pick up many mutations missed by Sanger sequencing. A recent review from Bert Vogelstein on Cancer Genome Landscapes reveals that ~140 genes can drive tumorigenesis that regulate the core processes of cell fate, cell survival, and genome maintenance. Most tumors will contain 2-8 of these, and the rest are passenger mutations. Focusing on driver mutations may help drive down sequencing and bioinformatics requirements. Wrapping up the talk, Dr. Bloom spoke about using the Illumina TruSight Tumor Panel for low frequency variant detection with high accuracy at Clarient’s SeqWright facility: “TruSight is a right-sized next-gen panel covering the key driver mutations, and is a cost effective and time efficient method for us”.

The next speaker had us all nervously contemplating what we’d eaten for breakfast. Steven Musser, from the FDA Center for Food Safety and Applied Nutrition discussed a pilot network of pathogen sequencing laboratories. Starting out with some alarming statistics Dr. Musser painted a picture of the problems with foodborne illness and difficulties tracking and tracing pathogens in the decentralized food supply in the United States. The public health need to speed up identification is obvious: the current method (PFGE) used to identify organisms and the source of contamination is often too late.  In order to intervene in outbreak situations, the time to identify pathogens must be compressed. Several outbreak studies featuring the particularly troublesome bugs Salmonella serotype Enteriditus and Salmonella Bareilly where next-generation sequencing clearly assisted identification and geographical spread were discussed. With regard to the role of NGS technology in food safety, Dr. Musser not surprisingly suggested that infrastructure, data generation and accession, and accuracy of accompanying metadata are considerations for its widespread and efficient adoption. He presented more information about GenomeTrakr, a project fueled by MiSeq that I touched on briefly in an earlier post. Using samples from  federal and state agencies, this effort will leverage genomic data to identify foodborne or unknown pathogens along with antibiotic resistance and virulence factors, and make this data easy to store and access in a central database. Although not quite ready for prime time yet, preliminary data from GenomeTrakr are looking very promising.

Next up, Mark Adams from the J. Craig Venter Institute discussed the challenges and opportunities in strain-level comparative genomics for organisms causing healthcare-associated infections (HAIs). An unintended and usually preventable healthcare consequence, HAIs affect ~1.7 million people and add billions of dollars to the cost of healthcare annually in the United States. Dr. Adams focused his talk around two particular organisms associated with HAIs, the emergent Acinetobacter baumannii and a more well-known culprit, Klebsiella pneumoniae.  A. baumannii was a HAI minor player that has recently skyrocketed cases of ventilator-associated pneumonia with the rapid emergence of widespread and drug resistant strains. This begs the question of what genomic adaptations are needed to become a successful pathogen. Using a comparative genomics approach in one Ohio hospital system, Dr. Adams sequenced 50 A. baumannii genomes using both short and long read technology. The study found that plasmid-chromosome interactions may lead to co-selection for resistance and virulence, and that redundant genetic mechanisms contribute to antibiotic resistance among the A. baumannii strains studied. Although long-read technology is very useful for hybrid assemblies, the accurate, high-resolution genomic information from HiSeq is critical for uncovering strain differences often masked by multi-locus sequence typing methods. The gram-negative Klebsiella pneumoniae is a normal member of our internal and external flora that can cause serious infections in immunocompromised patients, which is further complicated by antibiotic resistance. Dr. Adams presented an interesting screen based on GoldenGate genotyping to determine the presence or absence of resistance genes. Dr. Adams ended his presentation by emphasizing molecular diagnostic assays that predict resistance phenotypes in HAI organisms are needed to match the right drug to the right bug.

Just as our second or third cups of coffee were kicking in, it was time for a virology pop quiz from Michael Katze of the University of Washington. Pitching out some easy ones on the eight genes encoded by influenza and the precipitous dip in life expectancy caused by the 1918 pandemic, Dr. Katze got to an interesting question that had me stumped: what was the mortality rate of 1918 influenza? Multiple choice answers were 60%, 25%, and 2.5% and my hazarded guess was 25%. Turns out that the case fatality rate was ~2.5%, but the infection rate may have been as high as 1/3 of the world’s population, suggesting that host responses play a huge role in the success of viral infections. To discern these insights on how the host response is shaped by pathogenesis, Dr. Katze described his lab’s integrated approach to infectious disease using in vivo and in vitro models combined with functional proteomics, genomics, computational modeling, and biochemical approaches. This integrated, systems biology approach that his virology lab applies was a theme woven through a dizzying series of slides covering network analysis of H5N1 influenza, the SARS virus, and the emerging SARS-like novel coronavirus 5 from the Middle East (HCoV-EMC). This massive data integration effort revealed key immunological response differences to both types of SARS viruses, and offers some hints as to the pandemic potential of HCoV-EMC. Dr. Katze then took it into the third dimension with some amazing geometrical representations for visualizing similarities and differences between host response profiles. To say the very least, working in the Katze Lab must be an amazing cross-disciplinary experience, so check out a review in this month’s Nature Reviews Microbiology for the very latest on systems biology approaches.

George Weinstock from the Genome Institute at Washington University spoke next on emerging medical metagenomics. He illustrated the 2011 St Louis E. coli O157:H7 outbreak in supermarket prepared foods, where PFGE analysis of isolated strains was identical. Using a genomics approach and Backbone ORF SNP Set (BOSS) analysis, the SNPs were powerful enough to tie cases to the individual salad bars from which they’d picked up the infection. Dr. Weinstock next described the power of genomics with late onset bacteremia in the NICU. In this highly controlled environment where every diaper is stored in the freezer, they can examine the gut microbiome to track infections longitudinally. In combination with blood culture, they perform WGS followed by alignment and assembly, and cross check the results to determine the number of SNPs from self-alignment, providing sequencing or alignment errors versus the number of SNPs from cross-alignment, providing errors and variants. They find bacteremic strains frequently in the stool, and also patients who may be at risk for infection. With more information on high risk strains, it becomes possible to perform metagenomic analysis and skip the time-consuming process of direct culture. Another study from Peer Bork’s group in Heidelberg using a metagenomics shotgun approach in a different population of gut organisms provided evidence that the days of culturing an organism may indeed be on the decline; metagenomics analysis showed all present organisms and provided actionable results quickly. Dr. Weinstock finished with a comprehensive list of conditions and diseases that could benefit from metagenomics analysis, some of which are already being addressed by the Human Microbiome Project.

Finishing up the morning was a talk from Stephan Schuster of Penn State University about metagenomic and metatranscriptomic analyses of the complex microbial communities in wastewater. Citing the Roman aqueducts as the greatest of mankind’s inventions, Dr. Schuster described saturation sequencing of samples from a Singapore sewage plant. Looking at whole-genome and transcriptome sequencing, they found more than 2,000 genera, 10,000s of species, and scores of previously unclassified organisms. Key genes that remove nitrogen from wastewater were found in the low abundance bins of distribution, and would have been missed without deep sequencing. The picture for RNA was even more revealing, or perhaps revealing not much- over 85% of rRNAs have no annotation, opening up many more questions than answers. Metatranscriptome work from this sample alone will indeed be fertile ground for the work of future Ph.D. students for years to come.

All told, this small meeting represented an amazing repertoire of techniques and a fascinating look at the studies that genomic technology is enabling. A big thank you to all of the presenters, attendees, and organizers for making this Illumina Scientific Summit in the Big Easy a big success.

Image adapted from