Waibhav Tembe

Resolving Individuals Contributing Trace Amounts of DNA to Highly Complex Mixtures Using High-Density SNP Genotyping Microarrays

We use high-density single nucleotide polymorphism (SNP) genotyping microarrays to demonstrate the ability to accurately and robustly determine whether individuals are in a complex genomic DNA mixture. We first develop a theoretical framework for detecting an individuals presence within a mixture, then show, through simulations, the limits associated with our method, and finally demonstrate experimentally the identification of the presence of genomic DNA of specific individuals within a series of highly complex genomic mixtures, including mixtures where an individual contributes less than 0.1% of the total genomic DNA. These findings shift the perceived utility of SNPs for identifying individual trace contributors within a forensics mixture, and suggest future research efforts into assessing the viability of previously sub-optimal DNA sources due to sample contamination. These findings also suggest that composite statistics across cohorts, such as allele frequency or genotype counts, do not mask identity within genome-wide association studies. The implications of these findings are discussed.

Germline Mutations in HOXB13 and Prostate-Cancer Risk

BACKGROUND Family history is a significant risk factor for prostate cancer, although the molecular basis for this association is poorly understood. Linkage studies have implicated chromosome 17q21-22 as a possible location of a prostate-cancer susceptibility gene. METHODS We screened more than 200 genes in the 17q21-22 region by sequencing germline DNA from 94 unrelated patients with prostate cancer from families selected for linkage to the candidate region. We tested family members, additional case subjects, and control subjects to characterize the frequency of the identified mutations. RESULTS Probands from four families were discovered to have a rare but recurrent mutation (G84E) in HOXB13 (rs138213197), a homeobox transcription factor gene that is important in prostate development. All 18 men with prostate cancer and available DNA in these four families carried the mutation. The carrier rate of the G84E mutation was increased by a factor of approximately 20 in 5083 unrelated subjects of European descent who had prostate cancer, with the mutation found in 72 subjects (1.4%), as compared with 1 in 1401 control subjects (0.1%) (P=8.5x10(-7)). The mutation was significantly more common in men with early-onset, familial prostate cancer (3.1%) than in those with late-onset, nonfamilial prostate cancer (0.6%) (P=2.0x10(-6)). CONCLUSIONS The novel HOXB13 G84E variant is associated with a significantly increased risk of hereditary prostate cancer. Although the variant accounts for a small fraction of all prostate cancers, this finding has implications for prostate-cancer risk assessment and may provide new mechanistic insights into this common cancer. (Funded by the National Institutes of Health and others.).

Whole-genome sequencing of multiple myeloma from diagnosis to plasma cell leukemia reveals genomic initiating events, evolution, and clonal tides

The longitudinal evolution of a myeloma genome from diagnosis to plasma cell leukemia has not previously been reported. We used whole-genome sequencing (WGS) on 4 purified tumor samples and patient germline DNA drawn over a 5-year period in a t(4;14) multiple myeloma patient. Tumor samples were acquired at diagnosis, first relapse, second relapse, and end-stage secondary plasma cell leukemia (sPCL). In addition to the t(4;14), all tumor time points also shared 10 common single-nucleotide variants (SNVs) on WGS comprising shared initiating events. Interestingly, we observed genomic sequence variants that waxed and waned with time in progressive tumors, suggesting the presence of multiple independent, yet related, clones at diagnosis that rose and fell in dominance. Five newly acquired SNVs, including truncating mutations of RB1 and ZKSCAN3, were observed only in the final sPCL sample suggesting leukemic transformation events. This longitudinal WGS characterization of the natural history of a high-risk myeloma patient demonstrated tumor heterogeneity at diagnosis with shifting dominance of tumor clones over time and has also identified potential mutations contributing to myelomagenesis as well as transformation from myeloma to overt extramedullary disease such as sPCL.

Performance of feature-selection methods in the classification of high-dimension data

Contemporary biological technologies produce extremely high-dimensional data sets from which to design classifiers, with 20,000 or more potential features being common place. In addition, sample sizes tend to be small. In such settings, feature selection is an inevitable part of classifier design. Heretofore, there have been a number of comparative studies for feature selection, but they have either considered settings with much smaller dimensionality than those occurring in current bioinformatics applications or constrained their study to a few real data sets. This study compares some basic feature-selection methods in settings involving thousands of features, using both model-based synthetic data and real data. It defines distribution models involving different numbers of markers (useful features) versus non-markers (useless features) and different kinds of relations among the features. Under this framework, it evaluates the performances of feature-selection algorithms for different distribution models and classifiers. Both classification error and the number of discovered markers are computed. Although the results clearly show that none of the considered feature-selection methods performs best across all scenarios, there are some general trends relative to sample size and relations among the features. For instance, the classifier-independent univariate filter methods have similar trends. Filter methods such as the t-test have better or similar performance with wrapper methods for harder problems. This improved performance is usually accompanied with significant peaking. Wrapper methods have better performance when the sample size is sufficiently large. ReliefF, the classifier-independent multivariate filter method, has worse performance than univariate filter methods in most cases; however, ReliefF-based wrapper methods show performance similar to their t-test-based counterparts.

GRM7 variants confer susceptibility to age-related hearing impairment

Age-related hearing impairment (ARHI), or presbycusis, is the most prevalent sensory impairment in the elderly. ARHI is a complex disease caused by an interaction between environmental and genetic factors. Here we describe the results of the first whole genome association study for ARHI. The study was performed using 846 cases and 846 controls selected from 3434 individuals collected by eight centers in six European countries. DNA pools for cases and controls were allelotyped on the Affymetrix 500K GeneChip for each center separately. The 252 top-ranked single nucleotide polymorphisms (SNPs) identified in a non-Finnish European sample group (1332 samples) and the 177 top-ranked SNPs from a Finnish sample group (360 samples) were confirmed using individual genotyping. Subsequently, the 23 most interesting SNPs were individually genotyped in an independent European replication group (138 samples). This resulted in the identification of a highly significant and replicated SNP located in GRM7, the gene encoding metabotropic glutamate receptor type 7. Also in the Finnish sample group, two GRM7 SNPs were significant, albeit in a different region of the gene. As the Finnish are genetically distinct from the rest of the European population, this may be due to allelic heterogeneity. We performed histochemical studies in human and mouse and showed that mGluR7 is expressed in hair cells and in spiral ganglion cells of the inner ear. Together these data indicate that common alleles of GRM7 contribute to an individuals risk of developing ARHI, possibly through a mechanism of altered susceptibility to glutamate excitotoxicity.

Identification of extracellular miRNA in human cerebrospinal fluid by next-generation sequencing

There has been a growing interest in using next-generation sequencing (NGS) to profile extracellular small RNAs from the blood and cerebrospinal fluid (CSF) of patients with neurological diseases, CNS tumors, or traumatic brain injury for biomarker discovery. Small sample volumes and samples with low RNA abundance create challenges for downstream small RNA sequencing assays. Plasma, serum, and CSF contain low amounts of total RNA, of which small RNAs make up a fraction. The purpose of this study was to maximize RNA isolation from RNA-limited samples and apply these methods to profile the miRNA in human CSF by small RNA deep sequencing. We systematically tested RNA isolation efficiency using ten commercially available kits and compared their performance on human plasma samples. We used RiboGreen to quantify total RNA yield and custom TaqMan assays to determine the efficiency of small RNA isolation for each of the kits. We significantly increased the recovery of small RNA by repeating the aqueous extraction during the phenol-chloroform purification in the top performing kits. We subsequently used the methods with the highest small RNA yield to purify RNA from CSF and serum samples from the same individual. We then prepared small RNA sequencing libraries using Illuminas TruSeq sample preparation kit and sequenced the samples on the HiSeq 2000. Not surprisingly, we found that the miRNA expression profile of CSF is substantially different from that of serum. To our knowledge, this is the first time that the small RNA fraction from CSF has been profiled using next-generation sequencing.

Profiles of Extracellular miRNA in Cerebrospinal Fluid and Serum from Patients with Alzheimer's and Parkinson's Diseases Correlate with Disease Status and Features of Pathology

The discovery and reliable detection of markers for neurodegenerative diseases have been complicated by the inaccessibility of the diseased tissue- such as the inability to biopsy or test tissue from the central nervous system directly. RNAs originating from hard to access tissues, such as neurons within the brain and spinal cord, have the potential to get to the periphery where they can be detected non-invasively. The formation and extracellular release of microvesicles and RNA binding proteins have been found to carry RNA from cells of the central nervous system to the periphery and protect the RNA from degradation. Extracellular miRNAs detectable in peripheral circulation can provide information about cellular changes associated with human health and disease. In order to associate miRNA signals present in cell-free peripheral biofluids with neurodegenerative disease status of patients with Alzheimers and Parkinsons diseases, we assessed the miRNA content in cerebrospinal fluid and serum from postmortem subjects with full neuropathology evaluations. We profiled the miRNA content from 69 patients with Alzheimers disease, 67 with Parkinsons disease and 78 neurologically normal controls using next generation small RNA sequencing (NGS). We report the average abundance of each detected miRNA in cerebrospinal fluid and in serum and describe 13 novel miRNAs that were identified. We correlated changes in miRNA expression with aspects of disease severity such as Braak stage, dementia status, plaque and tangle densities, and the presence and severity of Lewy body pathology. Many of the differentially expressed miRNAs detected in peripheral cell-free cerebrospinal fluid and serum were previously reported in the literature to be deregulated in brain tissue from patients with neurodegenerative disease. These data indicate that extracellular miRNAs detectable in the cerebrospinal fluid and serum are reflective of cell-based changes in pathology and can be used to assess disease progression and therapeutic efficacy.

G-SQZ

SUMMARY Large volumes of data generated by high-throughput sequencing instruments present non-trivial challenges in data storage, content access and transfer. We present G-SQZ, a Huffman coding-based sequencing-reads-specific representation scheme that compresses data without altering the relative order. G-SQZ has achieved from 65% to 81% compression on benchmark datasets, and it allows selective access without scanning and decoding from start. This article focuses on describing the underlying encoding scheme and its software implementation, and a more theoretical problem of optimal compression is out of scope. The immediate practical benefits include reduced infrastructure and informatics costs in managing and analyzing large sequencing data. AVAILABILITY http://public.tgen.org/sqz. Academic/non-profit: SOURCE available at no cost under a non-open-source license by requesting from the web-site; Binary: available for direct download at no cost. For-Profit: Submit request for for-profit license from the web-site.

Toward precision medicine in glioblastoma: the promise and the challenges

Integrated sequencing strategies have provided a broader understanding of the genomic landscape and molecular classifications of multiple cancer types and have identified various therapeutic opportunities across cancer subsets. Despite pivotal advances in the characterization of genomic alterations in glioblastoma, targeted agents have shown minimal efficacy in clinical trials to date, and patient survival remains poor. In this review, we highlight potential reasons why targeting single alterations has yielded limited clinical efficacy in glioblastoma, focusing on issues of tumor heterogeneity and pharmacokinetic failure. We outline strategies to address these challenges in applying precision medicine to glioblastoma and the rationale for applying targeted combination therapy approaches that match genomic alterations with compounds accessible to the central nervous system.

Integrated Genomic and Epigenomic Analysis of Breast Cancer Brain Metastasis

The brain is a common site of metastatic disease in patients with breast cancer, which has few therapeutic options and dismal outcomes. The purpose of our study was to identify common and rare events that underlie breast cancer brain metastasis. We performed deep genomic profiling, which integrated gene copy number, gene expression and DNA methylation datasets on a collection of breast brain metastases. We identified frequent large chromosomal gains in 1q, 5p, 8q, 11q, and 20q and frequent broad-level deletions involving 8p, 17p, 21p and Xq. Frequently amplified and overexpressed genes included ATAD2, BRAF, DERL1, DNMTRB and NEK2A. The ATM, CRYAB and HSPB2 genes were commonly deleted and underexpressed. Knowledge mining revealed enrichment in cell cycle and G2/M transition pathways, which contained AURKA, AURKB and FOXM1. Using the PAM50 breast cancer intrinsic classifier, Luminal B, Her2+/ER negative, and basal-like tumors were identified as the most commonly represented breast cancer subtypes in our brain metastasis cohort. While overall methylation levels were increased in breast cancer brain metastasis, basal-like brain metastases were associated with significantly lower levels of methylation. Integrating DNA methylation data with gene expression revealed defects in cell migration and adhesion due to hypermethylation and downregulation of PENK, EDN3, and ITGAM. Hypomethylation and upregulation of KRT8 likely affects adhesion and permeability. Genomic and epigenomic profiling of breast brain metastasis has provided insight into the somatic events underlying this disease, which have potential in forming the basis of future therapeutic strategies.