Technologies to Enable Autonomous Detection for BioWatch: Ensuring Timely and Accurate Information for Public Health Officials: Workshop Summary (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

I State of the Art for Autonomous Detection Systems Using Genomic Sequencing John Chris Detter, Ph.D., and I. Gary Resnick, Ph.D. A white paper prepared for the June 25–26, 2013, workshop on Strate- gies for Cost-Effective and Flexible Biodetection Systems That Ensure Timely and Accurate Information for Public Health Officials, hosted by the Institute of Medicine’s Board on Health Sciences Policy and the Na- tional Research Council’s Board on Life Sciences. The authors are re- sponsible for the content of this article, which does not necessarily represent the views of the Institute of Medicine or the National Research Council. BACKGROUND The BioWatch Program at the Department of Homeland Security (DHS) was developed and is currently operating to provide warnings of aerosol attacks with biological threat agents. A select number of urban centers have had BioWatch deployed for a number of years on a round- the-clock basis. These deployments have provided a great amount of op- erational experience that indicates great need for technology improve- ment. While the current BioWatch capability provides an important risk mitigation capability against biological warfare and terrorist (BW/BT) threats; many operationally significant challenges remain to be ad- dressed. A brief description of the desired capability, with extant chal- lenges, is presented below to provide a framework for discussion of technical approaches employing sequencing to address current technology limitations. 197 

198 TECHNOLOGIES TO ENABLE AUTONOMOUS DETECTION FOR BIOWATCH Affordable Continuous Coverage of At-Risk Populations The population of the United States is dispersed across a large land- mass. People converge to great population densities at numerous venues for varying lengths of time (e.g., day/night workday cycles between cit- ies and suburbs and special-event gatherings). Therefore, BioWatch must cover a large geographic area and varied indoor structures (e.g., special event centers, office buildings, transit centers, and underground rail sys- tems). To achieve this in a sustainable manner the capital and annual op- erating costs of the system must be commensurate with the assessed relative risk from BW/BT and the myriad of other needs faced by local, state, and federal officials. The current BioWatch system uses field aerosol concentration and sample collection, followed by the transport of samples back to a central laboratory and laboratory analysis of the samples. BioWatch data out- puts are provided to a decision-making body for an integrated analysis prior to taking response actions. Major decreases in the resources re- quired for BioWatch as well as improvements in the overall efficiency and effectiveness can be achieved through technical advances that pro- vide for field in situ detection and identification (an autonomous de- ployed detector) to eliminate sample transport and laboratory analysis costs; amplification-free nucleic acid detection to decrease reagent costs; reagent-free detection to decrease reagent costs, eliminate the need for environmental engineering controls, and minimize the need for electrical power for the deployed autonomous detector; inexpensive analysis of agent recognition events within the autonomous detector to decrease sensor unit production cost and maintenance; and system mod- ularity to minimize technology refresh costs. Accuracy and Precision Supporting High-Regret Responses Surveillance derives its value by informing response management systems that have the potential for eliminating or mitigating the impacts of risks. Response options vary in efficacy, cost, and associated negative consequences. There are also negative impacts associated with false- positive system outputs. Therefore, the accuracy and precision of the BioWatch system have a profound impact on overall system perfor- mance, value, and sustainability. Operational experience with the current BioWatch system indicates a strong need for improved accuracy while maintaining robust precision. 

APPENDIX I 199 The great diversity of the microbial world coupled with the fact that only a fraction of this diversity has been identified and characterized has re- sulted in a significant number of “environmental positives” that have adversely impacted BioWatch performance. The fact that no aerosol at- tacks have been detected and that no impacts of such attacks have been observed provides little evidence for understanding the false-negative potential. This is greatly compounded by the inherent uncertainty of the biothreat. Responsiveness to Full Scope of Biological Warfare and Terrorist Threats The uncertainty associated with the biothreat (e.g., which agents will be encountered at which locations, when it will occur, how much will be delivered, and how it will be dispersed) provides great operational con- straints on BioWatch. A large number of detector units are needed to cover populations at risk, and the system should be responsive to all po- tential biothreat agents (including emerging, re-emerging, and engi- neered pathogens) that may be presented as aerosol threats. In addition, an indication of unique agent phenotypic characteristics is desirable in order to guide response decisions (e.g., whether the specific strain en- countered is responsive to a particular antibiotic). The current BioWatch system is focused on a specific set of patho- gens and provides some level of identification. To be fully responsive to the potential bioaerosol threat, the scope of agents addressed must be greatly increased and the cost of coverage must be drastically decreased. CURRENT STATE OF SEQUENCING Next-generation sequencing (NGS) platforms have remarkable per- formance specifications. Most of them produce very high quality data in an automated or semiautomated fashion. Some are small, benchtop mod- els, able to produce large amounts of data in a relatively short time and for a relatively low cost. The rapid advancement in NGS technologies will soon enable pathogen detection devices to rely on sequencing to provide a wealth of information about the environment in a cost-effective and timely manner. NGS technologies can be used to sequence almost any sample con- taining biological material, such as clinical (human, animal), environ- 

200 TECHNOLOGIES TO ENABLE AUTONOMOUS DETECTION FOR BIOWATCH mental (water, soil, surfaces, plants, etc.), or pure cultures of organisms of interest. Sequencing can include DNA, RNA (in the form of cDNA), or both, depending on the type of information needed. For example, DNA sequencing can reveal which organisms are present in a sample and some of their phenotypic characteristics (e.g., antibiotic resistance). However, RNA sequencing (RNAseq) must be used for RNA viruses. RNAseq is also used to study the transcriptional profile of organisms at specific time points, allowing for a better understanding of their metabol- ic activity and for identification of genes that play key roles in disease, genetic disorders, inflammatory response, cancer, and so on. NGS technologies require that DNA molecules are converted to NGS libraries. RNA is always converted to DNA first, as currently there are no direct RNA sequencing technologies (although the potential exists with the Pacific Biosciences (PacBio) real-time sequencer (RS), which is not evaluated here because of its very large footprint). Standard library preparation processes include DNA fragmentation and the addition of appropriate adapter molecules to the ends of DNA fragments. Adapters are unique DNA sequences (usually 30 to 60 base pairs long), which al- low sequencing to occur, and they can also incorporate barcodes (or indi- ces) that enable analysis of many samples in parallel. Depending on the application, library preparation methods take between 2 hours and 3 days. Once the libraries are prepared, each DNA fragment present in the library is clonally amplified before sequencing. This process and its de- gree of automation depend on the NGS platform. For example, the Illumina platforms use clustering (MiSeq is fully automated), whereas 454 and IonTorrent platforms use emulsion polymerase chain reaction (PCR) techniques (which will soon be mostly automated on IonTorrent). NGS produces vast amounts of data, which are output as reads. A read is a string of DNA nucleotides corresponding to the sequence of the original DNA or RNA molecule in the sample. Each NGS platform out- puts reads with three important characteristics for interpretation: read length, the number of reads, and their quality (fidelity). Read lengths vary from less than 50 base pairs (bps) to >3,000 bps, depending on the platform and sequencing kits. Read numbers vary from as few as 1.5 mil- lion to 4 billion per run. Reads can be evaluated independently, or they can be combined into much longer strings of DNA or RNA sequences using a variety of computational tools. Sequencing of any sample generally requires four steps: nucleic acid extraction (sample preparation), library preparation, sequencing, and data analysis. Figure I-1 outlines a standard NGS workflow. Sequencing can 

APPENDIX I 201 be performed on any one of the current NGS platforms, with each having different requirements and often their own unique set of data output and management. Table I-1 briefly lists the facilities, personnel, equipment, software, and reagents needed for each platform. For many applications, sequencing offers great advantages over the traditional methods. For example, in the field of pathogen detection, NGS can identify not only known organisms but also indications of nov- el, emerging, and engineered ones. Comparative analysis to known path- ogens, the presence of virulence genes, and recombinant engineering markers and phylogenetic placement would provide indication of novel threat agents. This is highly relevant, especially for rapidly evolving and highly diverse organisms, such as RNA viruses and Burkholderia spp. In addition, NGS does not require prior knowledge of pathogens present in a sample as do the traditional detection methods. Therefore, NGS shows promise as the ultimate pathogen detection tool. Other application areas in which NGS will play a significant role include pathogen characteriza- tion (strain typing, antibiotic resistance, etc.), bioforensics, biometrics, and biosurveillance. For NGS to succeed in all these applications, basic studies and databases that correlate genotype (genetic sequence of an organism) to phenotype (the behavior of an organism, such as its patho- genicity, transmissibility, resistance, etc.) are required. Without bio- informatic analysis, the data cannot be used to make these determinations and inferences. Sample preparation (isolation of nucleic acids) Preparation of a sequencing libraries Sequencing Computational analysis of sequencing reads FIGURE I-1 Overview of the next-generation sequencing process. 

202 TECHNOLOGIES TO ENABLE AUTONOMOUS DETECTION FOR BIOWATCH TABLE I-1 Characteristic Differences Between Sequencing Platforms Facilities and Software and Sequencer Personnela Reagents Equipment Hardware MiSeq– 68.6×56.5×52.3 One to two Instrument in- Library Illumina cm (W×D×H). individuals are cludes all soft- preparation and Total weight: needed for ware required sequencing run 57.2 kg. sample receipt through data reagents are avail- Suggested lab through generation. able through bench space sequence data Analytical tools Illumina and can about 1.5× that generation. are available from be ordered via size available One or two multiple sources phone or the Inter- prior to individuals are and can utilize net. installation. suggested for platforms from a Alternate library The laboratory data PC laptop preparation kits are must be interpretation. through a large offered through maintained at server system. multiple scientific 22±3ºC for supply vendors. proper functioning. Can run on standard electrical systems. Roche454 Full size One to two Includes All standard library Roche454 GS individuals are sufficient preparation and FLX+ needed for software for sequence run rea- Upper assembly sample receipt data gents can be pur-  74.3×69.8 through generation and a chased ×36.1 cm sequence data multitude of directly from (W×D×H) generation. tools exist for Roche and  Includes data analysis integrate well with an 82.5-cm One or two and assembly. the platform. monitor individuals are The full-size Orders can be Lower assembly suggested for sequencer is easily placed via  75.2×90.8 data well suited to phone or Internet. ×92.7 cm interpretation. computational (W×D×H) systems ranging Total weight: from large 242 kg. desktop PCs to Can run on servers. standard electri- cal systems. GS Junior GS Junior may 40×60×40 cm be analyzed (W×L×H) with less Total weight: 25 computational kg. power, however Can run on (as with standard analyzing any electrical NGS data on systems, also such a small 

APPENDIX I 203 able to operate system), the between 85 and time required 264 VAC. may be substantial. IonTorrent 61×51×53 cm One to two Includes a Library (W×D×H). individuals are server to preparation and Total weight: 30 needed for support primary sequencing run kits kg. sample receipt analysis through can be Operates best through output file purchased between 20 and sequence data generation and directly through 25ºC with a generation. variant calling. LifeTech, the ven- humidity of 40– One or two Additional dor (phone or In- 60%. individuals are processes can be ternet sales both Optimal eleva- suggested for handled through supported). tion is below data “apps” available Library 2,000 m (LANL interpretation. through the preparation has successfully plug-in store. reagents other than operated in- PGM supports those sold by strument at cloud-based LifeTech are also >2,200 m). processing, commonly used The instrument much of the and can be runs on 100- to analysis can be purchased from a 240-V power done from any variety of (standard Internet-linked scientific supply electrical portal, not re- companies system) and quiring on-site requires 35 to computational 45 PSI argon hardware. gas for operation. a Separate personnel required for sequence generation and sequence analysis due to the highly specific training required for each skill. NOTE: When comparing the current NGS technologies consisting of Illumina, IonTorrent, PacBio, and 454, each technology has its own strengths and weak- nesses, including the cost of sequencing. If sample and library preparation pro- cesses are excluded (they are relatively similar), the cost of sequencing a mega base pair (Mbp) of DNA on each sequencing platform is as follows: Illumina MiSeq, $0.13/Mbp; IonTorrent PGM, $0.57/Mbp; PacBio, $1.40/Mbp; and 454, $6.7/Mbp. NOTE: FLX = flexibility; GS = genome sequencer; LANL = Los Alamos Na- tional Laboratory; m = meters; NGS = next-generation sequencing; PGM = per- sonal genome machine; PSI = pounds per square inch; VAC = volts alternating currents. SOURCE: Adapted from 2013 Los Alamos National Laboratory sequencing report to Department of Defense. The promise of NGS cannot be realized without significant invest- ments in the analysis of the produced data, typically referred to as bioin- 

204 TECHNOLOGIES TO ENABLE AUTONOMOUS DETECTION FOR BIOWATCH formatics. Analysis of NGS data is highly specialized, depending on the types and detail desired from a particular analysis. For the identification and characterization of a pathogen, there are several levels of analysis possible, each with different requirements and able to reach more or less detailed conclusions. For the BioWatch application it may be of value to have two levels of analysis available. The first would be a simple percent match to selected microorganisms that is automatically performed. The second would be an in-depth genomic analysis that would be performed as required. In conclusion, NGS shows promise for improving current microbiol- ogy, molecular biology, and analytical biochemistry methods and for providing new data streams that will help us understand the current state of pathogens and anticipate future changes in the microbial world. OVERCOMING THE SHORTCOMINGS OF THE CURRENT BIOWATCH SYSTEM Next-generation sequencing will soon become the ultimate tool for pathogen detection and characterization in clinical and environmental samples. Until recently, NGS was a slow and costly process. However, it is becoming cost-competitive and sufficiently rapid for many applica- tions. Even though NGS is unlikely to replace the current rapid and port- able pathogen detection platforms in the next couple of years, in many cases it will provide actionable information faster than the rapid systems. This is mainly due to the comprehensive information provided by an or- ganism’s entire sequence versus a few selected segments of the genome. It is the only technology that can perform all of the following tasks in parallel from almost any sample: (1) detect all known pathogens, in- cluding viruses, bacteria, and protozoa; (2) identify emerging patho- gens, whether they have evolved naturally or been engineered; and (3) characterize the pathogens (e.g., determine antibiotic resistance or pathogenicity). Over the next 2 to 3 years NGS applications will likely help generate a world map displaying the real-time status of all infectious diseases. The data will be provided by a global network of interconnected facilities that use NGS platforms. Sequencing data, combined with the computational models of disease progression and easy visualization, will enable the ac- curate prediction and monitoring of disease spread and will reduce the effects on human lives and local economies. 

APPENDIX I 205 With the existing or forthcoming hardware and software upgrades, NGS technology will provide actionable information in 8 to 48 hours (including sample preparation, analysis, and interpretation), depending on the platform, the number of samples, and types of information needed. The simplest process includes detection of known pathogens and deter- mination of some of their features, such as antibiotic resistance. More complex processes will involve identification of novel pathogens in mixed samples (clinical or environmental samples such as BioWatch aerosol samples), prediction of their pathogenicity and susceptibility to antibiotics, vaccine efficacy, and matching their identities to pathogens that previously caused serious outbreaks. The major types of sequencing data can generally be obtained with three different pipelines, each providing different amounts and types of information, depending on the user's requirements (see Table H-2). It is the only technology that can perform all of the following tasks in parallel from almost any sample:  Sequencing provides complete genomic picture of all microor- ganisms, not dependent on a priori selection of a small number of agents.  Phenotype can be predicted from the sequence, providing re- sponse guidance, such as which antibiotics to use and whether existing diagnostic tests will work.  Organisms altered by genetic engineering can be identified, providing coverage for the engineered threat.  Previously unidentified pathogens can be presumptively identi- fied by comparative analysis, as was done with SARS and novel coronavirus. This and the previous point should decrease false- negative issues.  The more in-depth information provided by a draft sequence should give greater accuracy, decreasing false-positive issues and facilitating response decision.  Can sequencing be done in a reagent-free system? If so, it would eliminate the need for reagents and environmental control, which will radically decrease cost.  Sequencing can achieve single organism’s recognition, providing extreme sensitivity without compromising accuracy. Below, we will attempt to address many of the current and future needs of the BioWatch Program as we understand them in key areas im- 

206 TECHNOLOGIES TO ENABLE AUTONOMOUS DETECTION FOR BIOWATCH TABLE I-2 Overview of the Three Next-Generation Sequencing Pipelines for Pathogen Detection and Characterization Pipeline Description Actionable Information 1. Amplicon Rapid sequencing of very Identify and characterize known Sequencing small portions of pathogen pathogens, and some emerging genomes ones. Able to test hundreds of samples in parallel. 2. Pathogen identifi- Full sequencing of Identify and characterize known cation and charac- environmental and clinical and emerging pathogens, terization in mixed samples including bacteria, viruses, and samples protozoa. 3. Pure-culture (iso- Whole-genome Can identify sequences late) whole-genome sequencing of one associated with specific sequencing pathogen isolated from a outbreaks. Allows rapid sample and grown in the detection of the same pathogen lab in future outbreaks. SOURCE: Adapted from 2013 Los Alamos National Laboratory sequencing report to Department of Defense. portant to its mission. The information content generated from today’s sequencing technologies already provides the base of what BioWatch needs. The challenge is to engineer a fieldable pathogen-detection plat- form based on nucleic acid sequencing. Sensitivity There is high confidence that a robust data stream will be generated from any sample type. Successful sequencing has been achieved at the level of a single bacterium. In addition, metagenomics and ancient DNA sequencing have also been successfully used for identification of small amounts of microorganisms. One drawback of sequencing all of the DNA (and/or RNA) in a sample is that a small amount of pathogen DNA can be overwhelmed by the background, so clever enrichment strategies may significantly increase the sensitivity. For example, some of the cur- rent NGS platforms isolate and amplify individual nucleic acid segments in oil vesicles leading to clonal nucleic acid samples for analysis. Future generations of sequencing technologies are expected to achieve even bet- ter sensitivities. 

APPENDIX I 207 Specificity This is one of the greatest strengths of sequencing, which should eliminate most false-positive and false-negative results. Thousands of sequencing centers around the world are sequencing various environ- ments for many different reasons. These activities are generating a valu- able knowledge base that will keep growing at no expense to BioWatch, which will improve the specificity over time even further. Comparative analysis of novel strains of microorganisms will be robust and automated due to the availability of comprehensive microbial databases and analyti- cal algorithms. Size and Functionality This is currently an issue due to how the devices were built and to the audiences they have been built for. Currently the health care industry is driving the technology to a smaller, less complicated device for the benchtop in a diagnostic laboratory. With proper motivation and custom- ary engineering designs, an even smaller platform can be built (i.e., look at the progress Apple is making on a regular basis to miniaturize for their market). Commercial drivers will push the industry without much in- vestment from the BioWatch community. However, properly placed mo- tivation will drive the technology to where BioWatch needs it to be in a timely fashion. The unique needs of an autonomous detector (e.g., ex- tended mean-time-to-failure, multitier analyses outputs, remote data transmission, viability assessment) could be identified and shared with interested sequencing platform developers to establish collaborative re- search and development initiatives. Flexibility Multiplexing with standard protocols is a strength of the current sequencing platforms and will likely remain so for future devices. Com- putational adjustments allow for the ultimate flexibility in NGS. Identifi- cation from a mixed or complex sample (i.e., from a filter) is also a strength of sequencing, where one allows the sequencer to generate base- line data on everything that is in the sample and relies on automated bio- informatics tools to sort out the details. This is commonplace today, and capabilities will be even stronger as more data are generated and as com- putational advancements occur. Community or metagenomic analysis 

208 TECHNOLOGIES TO ENABLE AUTONOMOUS DETECTION FOR BIOWATCH will decrease the sample processing requirements as well as provide an additional characterization modality. Suspect segments of nucleic acid will be assessed in the context of all nucleic acid in the aerosol sample, providing the opportunity to detect culture media constituents not nor- mally found in ambient aerosols. More Than Red/Green Decision Making The information content of sequencing and analysis is so high that not only does one get identification down to the strain, but the confi- dence in this call is usually very high, even with today’s technology. The information content also includes identification of novel and engineered organisms by comparative and phylogenetic analysis. Traditional recom- binant vectors can also be targeted to detect engineered pathogens. Re- sults can be generated automatically from raw data using analytical tools and then presented at multiple levels of complexity for BioWatch tech- nical, scientific personnel, and public health decision makers. Measuring Timelines Speed and process flow are enhanced for sequencing because one can gather small amounts of target for sequencing and because sensitivity is high by nature. Databases can also be pared down to do rapid on-board detector analyses. Capabilities are currently in hand to develop two lev- els or modes of analysis (Defense Threat Reduction Agency [DTRA] is developing one currently called EDGE). One can imagine mode 1 in which rapid analysis and identification is done on board the detector via a laptop-sized platform which looks at a pared-down database of patho- gens and near neighbors. The second stage would involve the data being remotely ported to a larger comparative analysis server farm capable of doing a much more extensive analysis and confirmatory target identifica- tion. This greater level of analysis would support the BioWatch Actiona- ble Result assessment process involving multiple participants. Measuring Interval Sequence throughput can allow for large batches of samples to be processed regularly. Technology will allow for an engineering decision to be made as needed in device or analysis design. One will likely be able to choose between a few samples at a faster throughput and at a 

APPENDIX I 209 higher cost (rapid mode) or a multiplexed process that is slower and has higher content (detailed mode). In doing this today, one may use amplicon sequencing of fewer primer pairs in rapid mode versus many primer sets in detailed mode. In tomorrow’s version one could do the same with sequencing a few samples at a time in rapid mode versus the detailed mode where the device would run one large, highly multiplexed run per day. Cost Initial investment on sequencing is relatively high compared with some of the other detector platforms. The payoff is in the throughput, single-stage analysis and associated need of less manpower. Future ad- vancements will become even cheaper due to reagentless sequencing. Automation Sample preparation and sequencing are highly automatable. Howev- er, technical challenges exist due to the current reagent-based platforms. Next-generation, reagentless platforms will allow for a much more au- tomatable system. The health care industry is highly focused on this is- sue, and it will help drive the mean time between failures down substantially. Operation Environment Current technology is aimed at the clinical laboratory and a standard research laboratory setting. As markets drive the development of smaller and more fieldable devices, the upcoming next-generation systems will be much more adaptable to field autonomous outdoor detection. Again, reagentless-type devices will naturally progress in this direction. Overall, the health care and related technology industry has a desire to overcome many of the hurdles the BioWatch community is also focused on. These advances will happen naturally and at little to no expense to the BioWatch community. However, properly placed motiva- tional tools and collaborations from the BioWatch community to NGS industry leaders will promote a quicker advancement aimed more pre- cisely at BioWatch’s mission. Below we generally discuss the three phases we see the evolution being focused. 

210 TECHNOLOGIES TO ENABLE AUTONOMOUS DETECTION FOR BIOWATCH READINESS LEVELS The Institute of Medicine provides three readiness levels for a BioWatch autonomous biodetector against which to assess sequencing technology. These levels are as described below. It was assumed that the engineering, testing, and fielding of an autonomous detector would require at least 2 years after the required components (TRL 4) were available.  Tier 1: fully automated biodetection system, capable of 24/7/365 unattended outdoor and indoor operation, that will be at a tech- nology readiness level of TRL 6-plus by 2016.  Tier 2: similar requirements, but will not reach a TRL 6-plus level until sometime between 2016 and 2020.  Tier 3: technologies that have the potential of meeting or exceed- ing the BioWatch requirements, but a fully automated, TRL 6- plus system would take us beyond the 2020 time frame. For the- se technologies, describe the current critical paths (“long poles”) in meeting the BioWatch requirements and how they might be addressed. Tier 1 One way to achieve a Tier 1 system would be to expedite the engi- neering of an available NGS technology for high-throughput amplicon sequencing to create an autonomous field-deployable biodetector for the BioWatch system. This capability would allow inexpensive, rapid, and very detailed analysis of many samples in parallel. The parallel-processing capability could be used for analyzing multiple aerosol samples, providing narrow- er time cuts for sampling or numerous distinct amplicons. Currently available and validated primer sets can be easily utilized in this approach, but many others can be added. Amplicon sequencing can detect many known pathogens and their phenotypic features (antibiotic resistance markers and virulence factors). It can also detect and characterize some emerging pathogens, but this capability is limited. Summary of features:  Detects all known pathogens, antibiotic resistance markers, and virulence factors. 

APPENDIX I 211  Uses multiplexed end-point PCR to generate amplicons that are directly sequenced.  Can sequence hundreds or thousands of amplicons.  Has limited ability to identify emerging threats, especially RNA viruses. A fully automated platform should be smaller than 24 ft3, should use a regular 110-V outlet, and should automatically analyze samples every 8 hours following sample collection. Tier 2 A Tier 2 system could be achieved by using improved NGS technol- ogy to provide for metagenomic sequencing and analysis. This capability would sequence all nucleic acids present in a sample. It would detect and characterize not only the known pathogens but also most emerging ones. Because small amounts of a pathogen in a large amount of background would severely compromise the platform sensitivity, a strategy for re- moval of environmental (non-informative) nucleic acids may be desira- ble or required. Since all NGS platforms require library preparation, sample-to-result time would still be 6 to 10 hours following sample collection. Summary of features:  Detects all known pathogens, antibiotic resistance markers, and virulence factors.  Can detect many emerging pathogens, including engineered ones.  Even with improved sequencing speeds, the requirement to pre- pare libraries slows the overall process down. A fully automated platform should be smaller than 24 ft3, should use a regular 110-V outlet, and should automatically analyze samples every 8 hours. Tier 3 A Tier 3 system could be achieved through using a future sequencing technology (i.e., Oxford Nanopore or Gen2 PacBio) to rapidly perform metagenomic sequencing of environmental samples. This approach 

212 TECHNOLOGIES TO ENABLE AUTONOMOUS DETECTION FOR BIOWATCH would have similar capabilities to the Tier 2 platform, but the sample-to- result times would be much shorter—approximately 1 hour. No library preparation would be required, and the sequencing itself would be much more rapid. The soon-to-be-released Oxford Nanopore technology is ex- pected to offer such a technology advance. Summary of features:  Detects all known pathogens, antibiotic resistance markers, and virulence factors.  Can detect many emerging pathogens, including engineered ones. A fully automated platform should be smaller than 3 ft3, should use a regular 110-V outlet, and should automatically analyze a sample in 1 hour (in addition to air sampling, which can vary depending on the appli- cation). Reagent requirements should be minimal. SUMMARY Sequencing technology driven by strong and diverse markets has and will continue to make rapid advances. In addition, application of existing technologies is enabling rapid growth of genomic databases and filling in the microbial tree of life. The decrease in cost and increase in functional- ity of sequencing technology, coupled with publically available molecu- lar databases, should drive a growing interest in genomics for decades. Major points to consider in assessing the utility of sequencing to the BioWatch mission are as follows:  Great market pressure will facilitate advances in sequencing that can be adopted by BioWatch. In particular, research and devel- opment for capabilities to satisfy point-of-care and field analysis capabilities will drive development of core sensor components of direct value to BioWatch.  Tier 1 and Tier 2 timelines would benefit from comprehensive information provided by NGS, but the engineered autonomous detector would continue to require environmental controls and electrical power. Reagent costs could be decreased and have an analysis time of less than 10 hours following sample collection. Complexity of system fluidics would require routine maintenance. 

APPENDIX I 213  Tier 3 capabilities hold promise for eliminating reagents and minimizing the need for environmental controls. This would provide the benefits of information derived from sequencing while decreasing the costs associated with reagent systems. Again, a sequencing and analysis time of less than 10 hours fol- lowing sample collection should be possible.  An acquisition strategy for DHS should be considered that fos- ters research and development of specific technology compo- nents within the private sector, toward common desired capabilities. In summary, the inherent information content from sequencing and analysis should meet all the needs of BioWatch, addressing false-positive and false-negative issues. However, for at least the Tier 1 and Tier 2 timelines, significant engineering challenges will have to be overcome to adapt the existing reagent-based systems into an autonomous biodetector with desired attributes. Fortunately, extensive investments are being made by many public- and private-sector entities to develop technology that should provide the core sensor technology meeting the Tier 3 time- line and system requirements.