Informatics Challenges Research Articles

Implementing routine patient-reported outcome collection in a large, academic health system.

7053 Background: Patients' symptoms and side effects have traditionally been assessed by clinicians. There is increasing evidence that patient self-reported symptom severity often differs from clinician assessment, and that collecting patient-reported outcomes (PRO) can improve communication, symptom management, and even survival. However, the implementation of routine PRO collection across a large healthcare system poses operational and informatics challenges. Methods: Using native electronic health record (EHR) functionality, we implemented a standardized PRO questionnaire across a large academic cancer center and associated community-based practices. The questionnaire is based on the Patient-Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO-CTCAE) developed by the National Cancer Institute. It assesses eleven common side effects of cancer care and is available for completion from home via the EHR patient portal or in clinic via tablet PC. Implementation was stepwise, beginning with a disease-specific patient population in the main academic cancer center and expanding over two years to include all cancer types, three specialties (radiation oncology, hematology/oncology, gynecologic oncology), and multiple satellite practice locations. Results: PRO collection was initiated for patients with gastrointestinal malignancies in two clinic locations at the main cancer center in 12/2017. During the first 3 months of implementation (12/2017-2/2018), questionnaires were completed for 1838 (56.3%) of 3267 eligible patient visits. Work with practice managers and staff to refine operational workflows led to improvement to a 75.6% capture rate for the period 3/2018 – 5/2018. From 6/2018 through 6/2019, the program was expanded to all multidisciplinary clinics in the main cancer center, as well as eight satellite practices. Aggregate capture rates from 7/2018 through 12/2019 have shown sustained performance, with 101,082 (76.7%) of 131,720 eligible visits captured. Of twelve total clinics participating, eleven have sustained capture rates above 70%, and nine capture over 80% of eligible visits. Questionnaires were completed through the online patient portal 12.1% of the time, with the remainder completed in clinic via tablet PC. Conclusions: Routine PRO collection as standard-of-care is possible across a variety of practice environments in a large, complex health system, with sustained capture of approximately three-fourths of eligible visits. Most patients prefer to complete the questionnaire in clinic.

Automated abstraction of real-world clinical outcome in lung cancer: A natural language processing and artificial intelligence approach from electronic health records.

e14062 Background: Real world evidence generated from electronic health records (EHRs) is playing an increasing role in health care decisions. It has been recognized as an essential element to assess cancer outcomes in real-world settings. Automatically abstracting outcomes from notes is becoming a fundamental challenge in medical informatics. In this study, we aim to develop a system to automatically abstract outcomes (Progression, Response, Stable Disease) from notes in lung cancer. Methods: A lung cancer cohort (n = 5,003) was obtained from the Mount Sinai Data Warehouse. The progress, pathology and radiology notes of patients were used. We integrated various techniques of Natural Language Processing (NLP) and Artificial Intelligence (AI) and developed a system to automatically abstract outcomes. The corresponding images, biopsies and lines of treatments (LOTs) were abstracted as attributes of outcomes. This system includes four information models: 1. Customized NLP annotator model: preprocessor, section detector, sentence splitter, named entity recognition, relation detector; CRF and LSTM methods were applied to recognize entities and relations. 2. Clinical Outcome container model: biopsy evidence extractor, lines of treatment detector, image evidence extractor, clinical outcome event recognizer, date detector, and temporal reasoning; Domain-specific rules were crafted to automatically infer outcomes. 3. Document Summarizer; 4. Longitudinal Outcome Summarizer. Results: To evaluate the outcomes abstracted, we curated a subset (n = 792) from patient cohort for which LOTs were available. About 61% of the outcomes identified were supported by radiologic images (time window = ±14 days) or biopsy pathology results (time window = ±100 days). In 91% (720/792) of patients, Progression was abstracted within a time window of 90 days prior to first-line treatment. Also, 72% of the Progression events identified were accompanied by a downstream event (e.g., treatment change or death). We randomly selected 250 outcomes for manual curation, and 197 outcomes were assessed to be correct (precision = 79%). Moreover, our automated abstraction system improved human abstractor efficiency to curate outcomes, reducing curation time per patient by 90%. Conclusions: We have demonstrated the feasibility and effectiveness of NLP and AI approaches to abstract outcomes from lung cancer EHR data. It promises to automatically abstract outcomes and other clinical entities from notes across all cancers.

Quantitative Imaging Informatics for Cancer Research.

PURPOSEWe summarize Quantitative Imaging Informatics for Cancer Research (QIICR; U24 CA180918), one of the first projects funded by the National Cancer Institute (NCI) Informatics Technology for Cancer Research program.METHODSQIICR was motivated by the 3 use cases from the NCI Quantitative Imaging Network. 3D Slicer was selected as the platform for implementation of open-source quantitative imaging (QI) tools. Digital Imaging and Communications in Medicine (DICOM) was chosen for standardization of QI analysis outputs. Support of improved integration with community repositories focused on The Cancer Imaging Archive (TCIA). Priorities included improved capabilities of the standard, toolkits and tools, reference datasets, collaborations, and training and outreach.RESULTSFourteen new tools to support head and neck cancer, glioblastoma, and prostate cancer QI research were introduced and downloaded over 100,000 times. DICOM was amended, with over 40 correction proposals addressing QI needs. Reference implementations of the standard in a popular toolkit and standalone tools were introduced. Eight datasets exemplifying the application of the standard and tools were contributed. An open demonstration/connectathon was organized, attracting the participation of academic groups and commercial vendors. Integration of tools with TCIA was improved by implementing programmatic communication interface and by refining best practices for QI analysis results curation.CONCLUSIONTools, capabilities of the DICOM standard, and datasets we introduced found adoption and utility within the cancer imaging community. A collaborative approach is critical to addressing challenges in imaging informatics at the national and international levels. Numerous challenges remain in establishing and maintaining the infrastructure of analysis tools and standardized datasets for the imaging community. Ideas and technology developed by the QIICR project are contributing to the NCI Imaging Data Commons currently being developed.

PoleSat_2018: an optimized, automated, geomatics IT tool based on a gravitational model: strategic decision support in hospital catchment area planning

In France and elsewhere, decision-makers, healthcare professionals and health planners need to better understand and specify the provision of medical care. To this end, a hospital-based research project on a gravitational health planning modelling process was initiated in 2002. Since then, geomatics has emerged as a major scientific field for facing new challenges in medical informatics and health planning, thanks to the use of attractive interfaces, new methods and user-friendly IT technologies. Our initial 2002 model has recently been enhanced, optimized and automated as part of a spatial decision support system (PoleSat_2018). These decisive improvements and optimizations were mainly based on Delaunay triangulation, the replacement of human expertise with a heuristic dominance rule that provides a complete automated algorithm, and an online graphical user interface. Rapid, easy planning scenarios (by grouping and/or closing hospitals) give a quasi-instantaneous, strategic visualization of hospital catchment areas for decision-makers who are not experts in geomatics via ready-to-use maps and spreadsheets. This new implementation achieves our main objective, since the proposed deterministic method provides a completely automated, stable algorithm. A custom version of this tool is now being used by the French Ministry of Health for real planning issues. Consequently, PoleSat could be easily generalized as a prospective, strategic decision support tool for various health planning issues.

Open Science principles for accelerating trait-based science across the Tree of Life.

Synthesizing trait observations and knowledge across the Tree of Life remains a grand challenge for biodiversity science. Species traits are widely used in ecological and evolutionary science, and new data and methods have proliferated rapidly. Yet accessing and integrating disparate data sources remains a considerable challenge, slowing progress toward a global synthesis to integrate trait data across organisms. Trait science needs a vision for achieving global integration across all organisms. Here, we outline how the adoption of key Open Science principles-open data, open source and open methods-is transforming trait science, increasing transparency, democratizing access and accelerating global synthesis. To enhance widespread adoption of these principles, we introduce the Open Traits Network (OTN), a global, decentralized community welcoming all researchers and institutions pursuing the collaborative goal of standardizing and integrating trait data across organisms. We demonstrate how adherence to Open Science principles is key to the OTN community and outline five activities that can accelerate the synthesis of trait data across the Tree of Life, thereby facilitating rapid advances to address scientific inquiries and environmental issues. Lessons learned along the path to a global synthesis of trait data will provide a framework for addressing similarly complex data science and informatics challenges.

A negative storage model for precise but compact storage of genetic variation data.

Falling sequencing costs and large initiatives are resulting in increasing amounts of data available for investigator use. However, there are informatics challenges in being able to access genomic data. Performance and storage are well-appreciated issues, but precision is critical for meaningful analysis and interpretation of genomic data. There is an inherent accuracy vs. performance trade-off with existing solutions. The most common approach (Variant-only Storage Model, VOSM) stores only variant data. Systems must therefore assume that everything not variant is reference, sacrificing precision and potentially accuracy. A more complete model (Full Storage Model, FSM) would store the state of every base (variant, reference and missing) in the genome thereby sacrificing performance. A compressed variation of the FSM can store the state of contiguous regions of the genome as blocks (Block Storage Model, BLSM), much like the file-based gVCF model. We propose a novel approach by which this state is encoded such that both performance and accuracy are maintained. The Negative Storage Model (NSM) can store and retrieve precise genomic state from different sequencing sources, including clinical and whole exome sequencing panels. Reduced storage requirements are achieved by storing only the variant and missing states and inferring the reference state. We evaluate the performance characteristics of FSM, BLSM and NSM and demonstrate dramatic improvements in storage and performance using the NSM approach.

Running Genomic Analyses in the Cloud.

Genomic testing is rapidly moving into healthcare practice. However it comes with informatics challenges that the healthcare system has not previously faced - the raw data can be hundreds of gigabytes per test, the compute demands can be thousands of CPU hours, and the test can reveal deeply private health-srelated information that can have implications for anyone related to the person tested. While not a panacea, cloud computing has particular properties that can ameliorate some of these difficulties. This paper presents some of the key lessons learned while deploying a set of genomic analyses on cloud computing for Queensland Genomics.

Privacy-Preserving Predictive Modeling: Harmonization of Contextual Embeddings From Different Sources.

BackgroundData sharing has been a big challenge in biomedical informatics because of privacy concerns. Contextual embedding models have demonstrated a very strong representative capability to describe medical concepts (and their context), and they have shown promise as an alternative way to support deep-learning applications without the need to disclose original data. However, contextual embedding models acquired from individual hospitals cannot be directly combined because their embedding spaces are different, and naive pooling renders combined embeddings useless.ObjectiveThe aim of this study was to present a novel approach to address these issues and to promote sharing representation without sharing data. Without sacrificing privacy, we also aimed to build a global model from representations learned from local private data and synchronize information from multiple sources.MethodsWe propose a methodology that harmonizes different local contextual embeddings into a global model. We used Word2Vec to generate contextual embeddings from each source and Procrustes to fuse different vector models into one common space by using a list of corresponding pairs as anchor points. We performed prediction analysis with harmonized embeddings.ResultsWe used sequential medical events extracted from the Medical Information Mart for Intensive Care III database to evaluate the proposed methodology in predicting the next likely diagnosis of a new patient using either structured data or unstructured data. Under different experimental scenarios, we confirmed that the global model built from harmonized local models achieves a more accurate prediction than local models and global models built from naive pooling.ConclusionsSuch aggregation of local models using our unique harmonization can serve as the proxy for a global model, combining information from a wide range of institutions and information sources. It allows information unique to a certain hospital to become available to other sites, increasing the fluidity of information flow in health care.

Pharmacy informatics in multihospital health systems: Opportunities and challenges.

Challenges and opportunities in managing pharmacy-related technology in a multihospital health system are reviewed. With electronic medical record (EMR) implementations, pharmacy technology deployments, and increased numbers of hospitals merging into single health systems, opportunities and challenges for pharmacy informatics (PI) teams have grown. Pharmacy leaders must consider the implications of using technology in a multihospital health-system environment, as well as the impact of the health system's organizational structures on technology implementations and dedicated support teams. Common challenges in achieving EMR and other technology implementation and standardization initiatives in multihospital health systems include harmonization of practices across hospitals of various sizes and types and issues of platform compatibility and interoperability. PI teams must collaborate with information technology teams at the system level to identify practical strategies for making the best use of available resources to implement pharmacy automation and software to help pharmacists continue to provide safe and effective patient care. The organizational structures that affect informatics teams, pharmacy integration and standardization initiatives, formulary management practices, data management and analytics, and clinical decision support systems all must be areas of focus. An integrated pharmacy enterprise can be well positioned to leverage operational efficiencies gained from appropriate use of technology to enhance patient care. Careful attention must be paid to the manner in which these systems are designed, implemented, and managed in order to make the best use of the technological resources used by the health system.

Equality Challenges in the Use of eHealth: Selected Results from a Danish Citizens Survey.

The increased focus on and use of citizen/patient generated health data has spurred a wide range of personal health technology projects within digital intervention in health, e.g. telehealth. These developments are focused on objectives of improving health, but also claiming to provide health services at a lower cost. However, the question is: do the ways healthcare technologies are designed and developed support and improve healthcare services for those who need it the most? Survey data from our study point toward health informatics challenges in reaching the people who are considered in 'health risk' group, who, in this study, are interpreted as people with low level of education. The study shows that this group is less likely to use and communicate through health information technologies (HIT) and is generally more sceptical towards the benefits of HIT. We conclude that there is a need to pay specific attention to the patient groups that are socio-economically and health wise weakest during HIT design and development. It would also provide equality and equity in digital health intervention and access to healthcare for them in the future.

Mathematics, Informatics, Science and Education International Conference (MISEIC)

PrefaceThe Mathematics, Informatics, Science and Education International Conference (MISEIC) 2017 is organized by Universitas Negeri Surabaya is motivated to bridge the scientists, education experts and practicioners, and students in the scientific forum through sharing ideas and issues about theoretical and practical knowledge in mathematics, informatics, science and education. This seminar will be held on September 9, 2017 and will take place in Surabaya, Indonesia. The theme for this conference is “Building Togetherness for The Upcoming Challenges in Mathematics, Informatics, Science, and Education”. Thus, this conference has aims: (1) to bring together the scientists, engineers, researchers and practitioners, students, and civil society organization representatives in the scientific forum. (2) To share and to discuss theoretical and practical knowledge about innovation in applied science and engineering.This conference hosted by the Universitas Negeri Surabaya is considered to accommodate discussion among researchers in the field of mathematics, informatics, science and education as a scientific forum. Therefore, the invited speakers in this conference are expert in the field of mathematics, informatics, science and education.The MISEIC 2017 took place in Surabaya, Indonesia, on 9 September 2017 involving more than 300 papers are participated from various topics including pure and applied mathematics, science and technology innovation, computer science, and innovation in mathematics, science, and computer science education. There are 75 selected papers that go through a strict peer reviewed process, and these papers will be published in the present conference proceedings.We would like to thank the organizing committee and the members of reviewers for their kind assistance in reviewing the papers. We would also extend our best gratitude to keynote speakers for their invaluable contribution and worthwhile ideas shared in the conference.The EditorsYuni Sri Rahayu (Universitas Negeri Surabaya, Indonesia)Rooselyna Ekawati (Universitas Negeri Surabaya)Asep Bayu Dani Nandiyanto (Universitas Pendidikan Indonesia)Agung Lukito (Universitas Negeri Surabaya)Setya Chendra Wibawa (Universitas Negeri Surabaya)List of Conference Photographs, Committee Chairman, Technical Chairman and Scientific Committee are available in this PDF.

Development of a video-simulation instrument for assessing cognition in older adults

BackgroundCommonly used methods to assess cognition, such as direct observation, self-report, or neuropsychological testing, have significant limitations. Therefore, a novel tablet computer-based video simulation was created with the goal of being valid, reliable, and easy to administer. The design and implementation of the SIMBAC (Simulation-Based Assessment of Cognition) instrument is described in detail, as well as informatics “lessons learned” during development.ResultsThe software emulates 5 common instrumental activities of daily living (IADLs) and scores participants’ performance. The modules were chosen by a panel of geriatricians based on relevance to daily functioning and ability to be modeled electronically, and included facial recognition, pairing faces with the correct names, filling a pillbox, using an automated teller machine (ATM), and automatic renewal of a prescription using a telephone. Software development included three phases 1) a period of initial design and testing (alpha version), 2) pilot study with 10 cognitively normal and 10 cognitively impaired adults over the age of 60 (beta version), and 3) larger validation study with 162 older adults of mixed cognitive status (release version). Results of the pilot study are discussed in the context of refining the instrument; full results of the validation study are reported in a separate article. In both studies, SIMBAC reliably differentiated controls from persons with cognitive impairment, and performance was highly correlated with Mini Mental Status Examination (MMSE) score.Several informatics challenges emerged during software development, which are broadly relevant to the design and use of electronic assessment tools. Solutions to these issues, such as protection of subject privacy and safeguarding against data loss, are discussed in depth. Collection of fine-grained data (highly detailed information such as time spent reading directions and the number of taps on screen) is also considered.ConclusionsSIMBAC provides clinicians direct insight into whether subjects can successfully perform selected cognitively intensive activities essential for independent living and advances the field of cognitive assessment. Insight gained from the development process could inform other researchers who seek to develop software tools in health care.

Omics AnalySIs System for PRecision Oncology (OASISPRO): a web-based omics analysis tool for clinical phenotype prediction.

Precision oncology is an approach that accounts for individual differences to guide cancer management. Omics signatures have been shown to predict clinical traits for cancer patients. However, the vast amount of omics information poses an informatics challenge in systematically identifying patterns associated with health outcomes, and no general purpose data mining tool exists for physicians, medical researchers and citizen scientists without significant training in programming and bioinformatics. To bridge this gap, we built the Omics AnalySIs System for PRecision Oncology (OASISPRO), a web-based system to mine the quantitative omics information from The Cancer Genome Atlas (TCGA). This system effectively visualizes patients' clinical profiles, executes machine-learning algorithms of choice on the omics data and evaluates the prediction performance using held-out test sets. With this tool, we successfully identified genes strongly associated with tumor stage, and accurately predicted patients' survival outcomes in many cancer types, including adrenocortical carcinoma. By identifying the links between omics and clinical phenotypes, this system will facilitate omics studies on precision cancer medicine and contribute to establishing personalized cancer treatment plans. This web-based tool is available at http://tinyurl.com/oasispro; source codes are available at http://tinyurl.com/oasisproSourceCode. Supplementary data are available at Bioinformatics online.

Big data management challenges in health research-a literature review.

Big data management for information centralization (i.e. making data of interest findable) and integration (i.e. making related data connectable) in health research is a defining challenge in biomedical informatics. While essential to create a foundation for knowledge discovery, optimized solutions to deliver high-quality and easy-to-use information resources are not thoroughly explored. In this review, we identify the gaps between current data management approaches and the need for new capacity to manage big data generated in advanced health research. Focusing on these unmet needs and well-recognized problems, we introduce state-of-the-art concepts, approaches and technologies for data management from computing academia and industry to explore improvement solutions. We explain the potential and significance of these advances for biomedical informatics. In addition, we discuss specific issues that have a great impact on technical solutions for developing the next generation of digital products (tools and data) to facilitate the raw-data-to-knowledge process in health research.

Abstract 3567: Sequencing the human TCRβ repertoire on the Ion S5™ System

Abstract Next-generation Sequencing (NGS) is proving an important tool in increasing understanding of the human immune system, and thereby cancer immunology. αβ-T cells are the primary constituents of human cell-mediated adaptive immunity. The antigen specificity of each αβ-T cell is encoded in the 500-600 bp transcript encompassing the variable portion of the rearranged TCRα and TCRβ subunits, which can be read via NGS in a process termed repertoire sequencing. Until now, the main challenge the field faces is the lack of a technology that can provide a contiguous read of 600 bp to minimize the complexity of designing bias-prone primers and informatics challenges of stitching short reads. Here we leverage the long read capability of Ion 530™ chip to comprehensively sequence all three CDR domains of the TCRβ chain. The Ion 530™ chip offers greater than 15 M productive reads, allowing a multiplex of 2-4 samples with sufficient coverage for most repertoire profiling studies. Initial testing with Leukocyte total RNA demonstrates that this multiplex PCR assay produced repertoires that were much more similar to data derived from 5’RACE protocol than the commonly used BIOMED2 primer set. This result suggested that the use of long reads minimizes bias by allowing targeting of less variable regions. To further assess the performance of the assay, we designed a model system of 30 plasmid controls containing common human T-cell CDR3 sequences. Each plasmid was amplified individually and sequenced to confirm the detection of a single clonal population. Analytical sensitivity of the assay and accuracy of the accompanied analysis solution were further evaluated by spiking in plasmid concentrations from 10 pg to 0.0001 pg (5 million to 50 copies) in a background of 100 ng cDNA reverse transcribed from leukocyte total RNA. Results showed the assay offers linearity over 5 orders of magnitude of decreasing input concentration. In summary, we have demonstrated a NGS workflow for TCRβ sequencing that offers multiplex flexibility on Ion S5 with sample to answer in less than 48 hours. For Research Use Only. Not for use in diagnostic procedures. Citation Format: Denise S. Topacio-Hall, Tim Looney, Yongming Sun, Lauren Miller, Elizabeth Linch, Geoffrey Lowman, Lifeng Lin, Mark Andersen, Fiona Hyland, Ann Mongan. Sequencing the human TCRβ repertoire on the Ion S5™ System [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 3567. doi:10.1158/1538-7445.AM2017-3567

Physiological Informatics: Collection and Analyses of Data from Wearable Sensors and Smartphone for Healthcare.

Physiological data from wearable sensors and smartphone are accumulating rapidly, and this provides us the chance to collect dynamic and personalized information as phenotype to be integrated to genotype for the holistic understanding of complex diseases. This integration can be applied to early prediction and prevention of disease, therefore promoting the shifting of disease care tradition to the healthcare paradigm. In this chapter, we summarize the physiological signals which can be detected by wearable sensors, the sharing of the physiological big data, and the mining methods for the discovery of disease-associated patterns for personalized diagnosis and treatment. We discuss the challenges of physiological informatics about the storage, the standardization, the analyses, and the applications of the physiological data from the wearable sensors and smartphone. At last, we present our perspectives on the models for disentangling the complex relationship between early disease prediction and the mining of physiological phenotype data.

Urgent Challenges for Local Public Health Informatics.

Urgent Challenges for Local Public Health Informatics.

Telling plant species apart with DNA: from barcodes to genomes.

Land plants underpin a multitude of ecosystem functions, support human livelihoods and represent a critically important component of terrestrial biodiversity—yet many tens of thousands of species await discovery, and plant identification remains a substantial challenge, especially where material is juvenile, fragmented or processed. In this opinion article, we tackle two main topics. Firstly, we provide a short summary of the strengths and limitations of plant DNA barcoding for addressing these issues. Secondly, we discuss options for enhancing current plant barcodes, focusing on increasing discriminatory power via either gene capture of nuclear markers or genome skimming. The former has the advantage of establishing a defined set of target loci maximizing efficiency of sequencing effort, data storage and analysis. The challenge is developing a probe set for large numbers of nuclear markers that works over sufficient phylogenetic breadth. Genome skimming has the advantage of using existing protocols and being backward compatible with existing barcodes; and the depth of sequence coverage can be increased as sequencing costs fall. Its non-targeted nature does, however, present a major informatics challenge for upscaling to large sample sets.This article is part of the themed issue ‘From DNA barcodes to biomes’.

The biorepository portal toolkit: an honest brokered, modular service oriented software tool set for biospecimen-driven translational research.

BackgroundHigh throughput molecular sequencing and increased biospecimen variety have introduced significant informatics challenges for research biorepository infrastructures. We applied a modular system integration approach to develop an operational biorepository management system. This method enables aggregation of the clinical, specimen and genomic data collected for biorepository resources.MethodsWe introduce an electronic Honest Broker (eHB) and Biorepository Portal (BRP) open source project that, in tandem, allow for data integration while protecting patient privacy. This modular approach allows data and specimens to be associated with a biorepository subject at any time point asynchronously. This lowers the bar to develop new research projects based on scientific merit without institutional review for a proposal.ResultsBy facilitating the automated de-identification of specimen and associated clinical and genomic data we create a future proofed specimen set that can withstand new workflows and be connected to new associated information over time. Thus facilitating collaborative advanced genomic and tissue research.ConclusionsAs of Janurary of 2016 there are 23 unique protocols/patient cohorts being managed in the Biorepository Portal (BRP). There are over 4000 unique subject records in the electronic honest broker (eHB), over 30,000 specimens accessioned and 8 institutions participating in various biobanking activities using this tool kit. We specifically set out to build rich annotation of biospecimens with longitudinal clinical data; BRP/REDCap integration for multi-institutional repositories; EMR integration; further annotated specimens with genomic data specific to a domain; build application hooks for experiments at the specimen level integrated with analytic software; while protecting privacy per the Office of Civil Rights (OCR) and HIPAA.

Abstract 5276: Impact of duplicate removal on low frequency NGS somatic variant calling

Abstract Cancer genomic profiles created by analysis of targeted Next Generation Sequencing (NGS) panels is emerging as a powerful tool for making informed clinical decisions. Of the critical informatics challenges to address, accurate mutation calls and allele frequency estimations after accounting for PCR-mediated artifacts are debated. The process of sample preparation for NGS sequencing involves amplification by PCR. While PCR is relatively error-free, mistakes early in DNA synthesis can be compounded, driving detection of spurious mutations and having an adverse impact on clinical reporting. Previous reports have addressed the utility of detecting and removing PCR duplicate reads in Mendelian applications but have rarely examined its use with targeted NGS panels. We performed deduplication with 3 widely used tools (Samtools, Samblaster, and Picard) to understand sensitivity to call low frequency alleles and any impact on false positive/negative rates. Furthermore, we evaluated effects of duplicate removal on targeted panels of varying sizes and effects of sample matrices using replicates of 7 verified reference samples with several digitally confirmed alleles with frequencies ranging from 1-5%. It is not practical to perform deduplication on PCR-enriched panels, therefore, we assessed 3 different hybrid enrichment panels of varying size (387kb, 1.3Mb, and 54Mb). Deduplication by Picard resulted in a greater decrease in the mean depth for the smaller panels (32-59%) compared to Exome (15%), showing that higher molecular diversity lowers duplication rates. Uniformity (percent of ROI with depth within 20% of the mean depth) improved 6-18% after deduplication for the smaller panels, but only 1% for the Exome. Independent of panel size, about 32% of the total reads were marked as duplicates, reducing the power to call low frequency variants by 18%. Importantly, after added sequencing 95-96% of onco-specific variants were detected post-deduplication with a lower limit of detection of 3% compared to 2.5% pre-removal. For low-quality DNA samples we find no benefit in added sequencing for any panel. We also generally observed higher sensitivity (0% to 10% for SNV and -3% to 3% for indels) post deduplication. Molecular diversity also varies by sample type. Intact DNA show higher molecular diversity and lower duplication rates than degraded FFPE samples. We profiled mixed-quality FFPE samples (n = 85), good-quality fresh frozen samples (n = 3), and NA12878 (n = 1) on our smallest panel and noted duplication rates of 64±14%, 40±5% and 30% respectively. On average, deduplication cut the number of SNV calls by 17.4%, with the FFPE samples affected the most (2-88%). From our analysis we recommend performing deduplication during analysis of targeted panels. While we observed the most benefit for smaller panels with low uniformity, improved variant sensitivity was seen regardless of panel size. During experimental design, we advise a worksheet to guide deduplication decisions. Citation Format: Jeran Stratford, Gunjan Hariani, Jeff Jasper, Chad Brown, Wendell Jones, Victor J. Weigman. Impact of duplicate removal on low frequency NGS somatic variant calling. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 5276.