A genome sequence assembly provides the foundation for studies of genotypic and phenotypic variation, genome structure, and evolution of the target organism. In the past four decades, there has been a surge of new sequencing technologies, and with these developments, computational scientists have developed new algorithms to improve genome assembly. Here we discuss the relationship between sequencing technology improvements and assembly algorithm development and how these are applied to extend and improve human and nonhuman genome assemblies.

In this work, we present the Genome Modeling System (GMS), an analysis information management system capable of executing automated genome analysis pipelines at a massive scale. The GMS framework provides detailed tracking of samples and data coupled with reliable and repeatable analysis pipelines. The GMS also serves as a platform for bioinformatics development, allowing a large team to collaborate on data analysis, or an individual researcher to leverage the work of others effectively within its data management system. Rather than separating ad-hoc analysis from rigorous, reproducible pipelines, the GMS promotes systematic integration between the two. As a demonstration of the GMS, we performed an integrated analysis of whole genome, exome and transcriptome sequencing data from a breast cancer cell line (HCC1395) and matched lymphoblastoid line (HCC1395BL). These data are available for users to test the software, complete tutorials and develop novel GMS pipeline configurations. The GMS is available at https://github.com/genome/gms.

The sensitivity of massively-parallel sequencing has confirmed that most cancers are oligoclonal, with subpopulations of neoplastic cells harboring distinct mutations. A fine resolution view of this clonal architecture provides insight into tumor heterogeneity, evolution, and treatment response, all of which may have clinical implications. Single tumor analysis already contributes to understanding these phenomena. However, cryptic subclones are frequently revealed by additional patient samples (e.g., collected at relapse or following treatment), indicating that accurately characterizing a tumor requires analyzing multiple samples from the same patient. To address this need, we present SciClone, a computational method that identifies the number and genetic composition of subclones by analyzing the variant allele frequencies of somatic mutations. We use it to detect subclones in acute myeloid leukemia and breast cancer samples that, though present at disease onset, are not evident from a single primary tumor sample. By doing so, we can track tumor evolution and identify the spatial origins of cells resisting therapy.

Abstract Summary: Despite recent progress, computational tools that identify gene fusions from next-generation whole transcriptome sequencing data are often limited in accuracy and scalability. Here, we present a software package, BreakFusion that combines the strength of reference alignment followed by read-pair analysis and de novo assembly to achieve a good balance in sensitivity, specificity and computational efficiency. Availability: http://bioinformatics.mdanderson.org/main/BreakFusion Contact: [email protected]; [email protected] Supplementary information: Supplementary data are available at Bioinformatics online

MOTIVATION: The sequencing of tumors and their matched normals is frequently used to study the genetic composition of cancer. Despite this fact, there remains a dearth of available software tools designed to compare sequences in pairs of samples and identify sites that are likely to be unique to one sample. RESULTS: In this article, we describe the mathematical basis of our SomaticSniper software for comparing tumor and normal pairs. We estimate its sensitivity and precision, and present several common sources of error resulting in miscalls. AVAILABILITY AND IMPLEMENTATION: Binaries are freely available for download at http://gmt.genome.wustl.edu/somatic-sniper/current/, implemented in C and supported on Linux and Mac OS X. CONTACT: [email protected]; [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

MOTIVATION: The expansion of cancer genome sequencing continues to stimulate development of analytical tools for inferring relationships between somatic changes and tumor development. Pathway associations are especially consequential, but existing algorithms are demonstrably inadequate. METHODS: Here, we propose the PathScan significance test for the scenario where pathway mutations collectively contribute to tumor development. Its design addresses two aspects that established methods neglect. First, we account for variations in gene length and the consequent differences in their mutation probabilities under the standard null hypothesis of random mutation. The associated spike in computational effort is mitigated by accurate convolution-based approximation. Second, we combine individual probabilities into a multiple-sample value using Fisher-Lancaster theory, thereby improving differentiation between a few highly mutated genes and many genes having only a few mutations apiece. We investigate accuracy, computational effort and power, reporting acceptable performance for each. RESULTS: As an example calculation, we re-analyze KEGG-based lung adenocarcinoma pathway mutations from the Tumor Sequencing Project. Our test recapitulates the most significant pathways and finds that others for which the original test battery was inconclusive are not actually significant. It also identifies the focal adhesion pathway as being significantly mutated, a finding consistent with earlier studies. We also expand this analysis to other databases: Reactome, BioCarta, Pfam, PID and SMART, finding additional hits in ErbB and EPHA signaling pathways and regulation of telomerase. All have implications and plausible mechanistic roles in cancer. Finally, we discuss aspects of extending the method to integrate gene-specific background rates and other types of genetic anomalies. AVAILABILITY: PathScan is implemented in Perl and is available from the Genome Institute at: http://genome.wustl.edu/software/pathscan.

Massively parallel sequencing technologies continue to alter the study of human genetics. As the cost of sequencing declines, next-generation sequencing (NGS) instruments and datasets will become increasingly accessible to the wider research community. Investigators are understandably eager to harness the power of these new technologies. Sequencing human genomes on these platforms, however, presents numerous production and bioinformatics challenges. Production issues like sample contamination, library chimaeras and variable run quality have become increasingly problematic in the transition from technology development lab to production floor. Analysis of NGS data, too, remains challenging, particularly given the short-read lengths (35-250 bp) and sheer volume of data. The development of streamlined, highly automated pipelines for data analysis is critical for transition from technology adoption to accelerated research and publication. This review aims to describe the state of current NGS technologies, as well as the strategies that enable NGS users to characterize the full spectrum of DNA sequence variation in humans.

MOTIVATION: DNA copy number aberration (CNA) is a hallmark of genomic abnormality in tumor cells. Recurrent CNA (RCNA) occurs in multiple cancer samples across the same chromosomal region and has greater implication in tumorigenesis. Current commonly used methods for RCNA identification require CNA calling for individual samples before cross-sample analysis. This two-step strategy may result in a heavy computational burden, as well as a loss of the overall statistical power due to segmentation and discretization of individual sample's data. We propose a population-based approach for RCNA detection with no need of single-sample analysis, which is statistically powerful, computationally efficient and particularly suitable for high-resolution and large-population studies. RESULTS: Our approach, correlation matrix diagonal segmentation (CMDS), identifies RCNAs based on a between-chromosomal-site correlation analysis. Directly using the raw intensity ratio data from all samples and adopting a diagonal transformation strategy, CMDS substantially reduces computational burden and can obtain results very quickly from large datasets. Our simulation indicates that the statistical power of CMDS is higher than that of single-sample CNA calling based two-step approaches. We applied CMDS to two real datasets of lung cancer and brain cancer from Affymetrix and Illumina array platforms, respectively, and successfully identified known regions of CNA associated with EGFR, KRAS and other important oncogenes. CMDS provides a fast, powerful and easily implemented tool for the RCNA analysis of large-scale data from cancer genomes.

Abstract Summary: Massively parallel sequencing technologies hold incredible promise for the study of DNA sequence variation, particularly the identification of variants affecting human disease. The unprecedented throughput and relatively short read lengths of Roche/454, Illumina/Solexa, and other platforms have spurred development of a new generation of sequence alignment algorithms. Yet detection of sequence variants based on short read alignments remains challenging, and most currently available tools are limited to a single platform or aligner type. We present VarScan, an open source tool for variant detection that is compatible with several short read aligners. We demonstrate VarScan's ability to detect SNPs and indels with high sensitivity and specificity, in both Roche/454 sequencing of individuals and deep Illumina/Solexa sequencing of pooled samples. Availability and Implementation: Source code and documentation freely available at http://genome.wustl.edu/tools/cancer-genomics implemented as a Perl package and supported on Linux/UNIX, MS Windows and Mac OSX. Contact: [email protected] Supplementary information: Supplementary data are available at Bioinformatics online.

BACKGROUND: DNA sequencing is now emerging as an important component in biomedical studies of diseases like cancer. Short-read, highly parallel sequencing instruments are expected to be used heavily for such projects, but many design specifications have yet to be conclusively established. Perhaps the most fundamental of these is the redundancy required to detect sequence variations, which bears directly upon genomic coverage and the consequent resolving power for discerning somatic mutations. RESULTS: We address the medical sequencing coverage problem via an extension of the standard mathematical theory of haploid coverage. The expected diploid multi-fold coverage, as well as its generalization for aneuploidy are derived and these expressions can be readily evaluated for any project. The resulting theory is used as a scaling law to calibrate performance to that of standard BAC sequencing at 8x to 10x redundancy, i.e. for expected coverages that exceed 99% of the unique sequence. A differential strategy is formalized for tumor/normal studies wherein tumor samples are sequenced more deeply than normal ones. In particular, both tumor alleles should be detected at least twice, while both normal alleles are detected at least once. Our theory predicts these requirements can be met for tumor and normal redundancies of approximately 26x and 21x, respectively. We explain why these values do not differ by a factor of 2, as might intuitively be expected. Future technology developments should prompt even deeper sequencing of tumors, but the 21x value for normal samples is essentially a constant. CONCLUSION: Given the assumptions of standard coverage theory, our model gives pragmatic estimates for required redundancy. The differential strategy should be an efficient means of identifying potential somatic mutations for further study.

BACKGROUND: Investigators in the biological sciences continue to exploit laboratory automation methods and have dramatically increased the rates at which they can generate data. In many environments, the methods themselves also evolve in a rapid and fluid manner. These observations point to the importance of robust information management systems in the modern laboratory. Designing and implementing such systems is non-trivial and it appears that in many cases a database project ultimately proves unserviceable. RESULTS: We describe a general modeling framework for laboratory data and its implementation as an information management system. The model utilizes several abstraction techniques, focusing especially on the concepts of inheritance and meta-data. Traditional approaches commingle event-oriented data with regular entity data in ad hoc ways. Instead, we define distinct regular entity and event schemas, but fully integrate these via a standardized interface. The design allows straightforward definition of a "processing pipeline" as a sequence of events, obviating the need for separate workflow management systems. A layer above the event-oriented schema integrates events into a workflow by defining "processing directives", which act as automated project managers of items in the system. Directives can be added or modified in an almost trivial fashion, i.e., without the need for schema modification or re-certification of applications. Association between regular entities and events is managed via simple "many-to-many" relationships. We describe the programming interface, as well as techniques for handling input/output, process control, and state transitions. CONCLUSION: The implementation described here has served as the Washington University Genome Sequencing Center's primary information system for several years. It handles all transactions underlying a throughput rate of about 9 million sequencing reactions of various kinds per month and has handily weathered a number of major pipeline reconfigurations. The basic data model can be readily adapted to other high-volume processing environments.