Lee Lichtenstein

Mutational heterogeneity in cancer and the search for new cancer-associated genes.

Major international projects are underway that are aimed at creating a comprehensive catalogue of all the genes responsible for the initiation and progression of cancer. These studies involve the sequencing of matched tumour–normal samples followed by mathematical analysis to identify those genes in which mutations occur more frequently than expected by random chance. Here we describe a fundamental problem with cancer genome studies: as the sample size increases, the list of putatively significant genes produced by current analytical methods burgeons into the hundreds. The list includes many implausible genes (such as those encoding olfactory receptors and the muscle protein titin), suggesting extensive false-positive findings that overshadow true driver events. We show that this problem stems largely from mutational heterogeneity and provide a novel analytical methodology, MutSigCV, for resolving the problem. We apply MutSigCV to exome sequences from 3,083 tumour–normal pairs and discover extraordinary variation in mutation frequency and spectrum within cancer types, which sheds light on mutational processes and disease aetiology, and in mutation frequency across the genome, which is strongly correlated with DNA replication timing and also with transcriptional activity. By incorporating mutational heterogeneity into the analyses, MutSigCV is able to eliminate most of the apparent artefactual findings and enable the identification of genes truly associated with cancer.

Landscape of genomic alterations in cervical carcinomas

Cervical cancer is responsible for 10–15% of cancer-related deaths in women worldwide. The aetiological role of infection with high-risk human papilloma viruses (HPVs) in cervical carcinomas is well established. Previous studies have also implicated somatic mutations in PIK3CA, PTEN, TP53, STK11 and KRAS as well as several copy-number alterations in the pathogenesis of cervical carcinomas. Here we report whole-exome sequencing analysis of 115 cervical carcinoma–normal paired samples, transcriptome sequencing of 79 cases and whole-genome sequencing of 14 tumour–normal pairs. Previously unknown somatic mutations in 79 primary squamous cell carcinomas include recurrent E322K substitutions in the MAPK1 gene (8%), inactivating mutations in the HLA-B gene (9%), and mutations in EP300 (16%), FBXW7 (15%), NFE2L2 (4%), TP53 (5%) and ERBB2 (6%). We also observe somatic ELF3 (13%) and CBFB (8%) mutations in 24 adenocarcinomas. Squamous cell carcinomas have higher frequencies of somatic nucleotide substitutions occurring at cytosines preceded by thymines (Tp*C sites) than adenocarcinomas. Gene expression levels at HPV integration sites were statistically significantly higher in tumours with HPV integration compared with expression of the same genes in tumours without viral integration at the same site. These data demonstrate several recurrent genomic alterations in cervical carcinomas that suggest new strategies to combat this disease.

Discovery and characterization of artifactual mutations in deep coverage targeted capture sequencing data due to oxidative DNA damage during sample preparation

As researchers begin probing deep coverage sequencing data for increasingly rare mutations and subclonal events, the fidelity of next generation sequencing (NGS) laboratory methods will become increasingly critical. Although error rates for sequencing and polymerase chain reaction (PCR) are well documented, the effects that DNA extraction and other library preparation steps could have on downstream sequence integrity have not been thoroughly evaluated. Here, we describe the discovery of novel C > A/G > T transversion artifacts found at low allelic fractions in targeted capture data. Characteristics such as sequencer read orientation and presence in both tumor and normal samples strongly indicated a non-biological mechanism. We identified the source as oxidation of DNA during acoustic shearing in samples containing reactive contaminants from the extraction process. We show generation of 8-oxoguanine (8-oxoG) lesions during DNA shearing, present analysis tools to detect oxidation in sequencing data and suggest methods to reduce DNA oxidation through the introduction of antioxidants. Further, informatics methods are presented to confidently filter these artifacts from sequencing data sets. Though only seen in a low percentage of reads in affected samples, such artifacts could have profoundly deleterious effects on the ability to confidently call rare mutations, and eliminating other possible sources of artifacts should become a priority for the research community.

Oncotator: cancer variant annotation tool.

Oncotator is a tool for annotating genomic point mutations and short nucleotide insertions/deletions (indels) with variant‐ and gene‐centric information relevant to cancer researchers. This information is drawn from 14 different publicly available resources that have been pooled and indexed, and we provide an extensible framework to add additional data sources. Annotations linked to variants range from basic information, such as gene names and functional classification (e.g. missense), to cancer‐specific data from resources such as the Catalogue of Somatic Mutations in Cancer (COSMIC), the Cancer Gene Census, and The Cancer Genome Atlas (TCGA). For local use, Oncotator is freely available as a python module hosted on Github (https://github.com/broadinstitute/oncotator). Furthermore, Oncotator is also available as a web service and web application at http://www.broadinstitute.org/oncotator/.

A feedback-controlled interface for treadmill locomotion in virtual environments

Virtual environments (VEs) allow safe, repeatable, and controlled evaluations of obstacle avoidance and navigation performance of people with visual impairments using visual aids. Proper simulation of mobility in a VE requires an interface, which allows subjects to set their walking pace. Using conventional treadmills, the subject can change their walking speed by pushing the tread with their feet, while leveraging handrails or ropes (self-propelled mode). We developed a feedback-controlled locomotion interface that allows the VE workstation to control the speed of the treadmill, based on the position of the user. The position and speed information is also used to implement automated safety measures, so that the treadmill can be halted in case of erratic behavior. We compared the feedback-controlled to the self-propelled mode by using speed-matching tasks (follow a moving object or match the speed of an independently moving scene) to measure the efficacy of each mode in maintaining constant subject position, subject control of the treadmill, and subject pulse rates. In addition, we measured the perception of speed in the VE on each mode. The feedback-controlled mode required less physical exertion than self-propelled. The average position of subjects on the feedback-controlled treadmill was always within a centimeter of the desired position. There was a smaller standard deviation in subject position when using the self-propelled mode than when using the feedback-controlled mode, but the difference averaged less than 6 cm across all subjects walking at a constant speed. Although all subjects underestimated the speed of an independently moving scene at higher speeds, their estimates were more accurate when using the feedback-controlled treadmill than the self-propelled.

Collision judgment when viewing minified images through a HMD visual field expander

Purpose: Patients with tunnel vision have great difficulties in mobility. We have developed an augmented vision head mounted device, which can provide patients 5x expanded field by superimposing minified edge images of a wider field captured by a miniature video camera over the natural view seen through the display. In the minified display, objects appear closer to the heading direction than they really are. This might cause users to overestimate collision risks, and therefore to perform unnecessary obstacle-avoidance maneuvers. A study was conducted in a virtual environment to test the impact of minified view on collision judgment. Methods: Simulated scenes were presented to subjects as if they were walking in a shopping mall corridor. Subjects reported whether they would make any contact with stationary obstacles that appeared at variable distances from their walking path. Perceived safe passing distance (PSPD) was calculated by finding the transition point from reports of yes to no. Decision uncertainty was quantified by the sharpness of the transition. Collision envelope (CE) size was calculated by summing up PSPD for left and right sides. Ten normally sighted subjects were tested (1) when not using the device and with one eye patched, and (2) when the see-through view of device was blocked and only minified images were visible. Results: The use of the 5x minification device caused only an 18% increase of CE (13cm, p=0.048). Significant impact of the device on judgment uncertainty was not found (p=0.089). Conclusion: Minification had only a small impact on collision judgment. This supports the use of such a minifying device as an effective field expander for patients with tunnel vision.

P-27: Maintaining Position and Display Perspective in a Walking Simulator while Self-Pacing on a Treadmill

We developed a feedback controller for a walking simulator composed of a treadmill and a rear projection screen. The controller keeps the subject centered and visual extent consistent across subjects and throughout individual trials while allowing subjects to set their own walking pace.

Abstract 3580: GATK CNV: copy-number variation discovery from coverage data

We propose and evaluate a novel algorithm for inferring germline and somatic copy number variation from whole exome sequencing (WES) and whole genome sequencing (WGS) data. Starting with the depth of aligned short reads from a cohort of samples, we use a Bayesian model for learning sequencing bias and simultaneously detecting CNV events using a hidden Markov model for change-point detection. A unified framework is used to call both germline and somatic CNVs. Denoising and event discovery are performed self-consistently to achieve maximum accuracy. In contrast to previous methods, our model naturally accounts for mixed sex cohorts and can detect events on sex chromosomes. Furthermore, we can detect excessively noisy samples and extract useful information within a probabilistic framework. Our implementation can also utilize Spark clusters, enabling the processing of larger cohorts and allowing for improved runtime performance. We benchmark the new method for precision, recall, and reproducibility of both germline and somatic variants. Evaluations are performed on a cohort of WES samples from The Cancer Genome Atlas with matching WGS data. For germline variants, we use blood normal samples and compare our calls on WES data against Genome STRiP calls on WGS data. We find that GATK CNV yields remarkably higher precision and recall compared to XHMM and CODEX software packages. For somatic variants, we compare our calls against TITAN and find a remarkably high concordance. Citation Format: Mehrtash Babadi, David I. Benjamin, Samuel K. Lee, Andrey Smirnov, Aaron Chevalier, Lee Lichtenstein, Valentin Ruano Rubio. GATK CNV: copy-number variation discovery from coverage data [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 3580. doi:10.1158/1538-7445.AM2017-3580

Biomimetic smart sensors for autonomous robotic behavior I: acoustic processing

Robotics are rapidly becoming an integral tool on the battlefield and in homeland security, replacing humans in hazardous conditions. To enhance the effectiveness of robotic assets and their interaction with human operators, smart sensors are required to give more autonomous function to robotic platforms. Biologically inspired sensors are an essential part of this development of autonomous behavior and can increase both capability and performance of robotic systems. Smart, biologically inspired acoustic sensors have the potential to extend autonomous capabilities of robotic platforms to include sniper detection, vehicle tracking, personnel detection, and general acoustic monitoring. The key to enabling these capabilities is biomimetic acoustic processing using a time domain processing method based on the neural structures of the mammalian auditory system. These biologically inspired algorithms replicate the extremely adaptive processing of the auditory system yielding high sensitivity over broad dynamic range. The algorithms provide tremendous robustness in noisy and echoic spaces; properties necessary for autonomous function in real world acoustic environments. These biomimetic acoustic algorithms also provide highly accurate localization of both persistent and transient sounds over a wide frequency range, using baselines on the order of only inches. A specialized smart sensor has been developed to interface with an iRobot Packbot® platform specifically to enhance its autonomous behaviors in response to personnel and gunfire. The low power, highly parallel biomimetic processor, in conjunction with a biomimetic vestibular system (discussed in the companion paper), has shown the systems autonomous response to gunfire in complicated acoustic environments to be highly effective.

Abstract 3641: ReCapSeg: Validation of somatic copy number alterations for CLIA whole exome sequencing

The presence of specific somatic copy number alterations (SCNAs) in tumor genomes can be used to predict sensitivity to specific treatments and to predict outcomes. CapSeg Revised (ReCapSeg) detects SCNAs from tumor sequencing data. The Broad Institute is integrating ReCapSeg into the Clinical Research Sequencing Platform (CRSP), a CLIA-certified platform, to enable physicians to use detected SCNAs in treatment decisions. By using sequencing data CRSP can produce SCNA reports using the same data being generated for other reports (e.g. somatic single nucleotide variants). ReCapSeg produces copy ratio estimates for regions of the genome and includes a caller that labels regions as amplified, deleted, or neutral. ReCapSeg leverages a panel of normals (PoN), which obviates the need for the matched normal for a given tumor sample. Validation for ReCapSeg, in CRSP, is repeatable and standardized with a process for creating a PoN and gathering performance metrics from case samples. Additional QC metrics and criteria on input sequencing data were identified and included, so that reports can indicate when reduced performance can be expected. Citation Format: Lee Lichtenstein, Betty Woolf, Alyssa MacBeth, Ozge Birsoy, Niall Lennon. ReCapSeg: Validation of somatic copy number alterations for CLIA whole exome sequencing. [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 3641.