Analyses

Varun Aggarwala and Benjamin Voight analyze human polymorphism data and develop an expanded sequence context model that explains >81% of variability in substitution probabilities, highlighting mutation-promoting motifs. Using their model, they present substitution intolerance scores for genes and a new intolerance score for amino acids, and demonstrate clinical use of the model in neuropsychiatric diseases.

Shamil Sunyaev, Alexander Gimelbrant and colleagues report an analysis of the genetic variability in human monoallelically expressed genes. They find that genes with monoallelic expression show greater genetic diversity than biallelically expressed genes and that this diversity is associated with greater allelic age.

Andrea Sottoriva, Trevor Graham and colleagues analyze tumor sequencing data and show that a substantial proportion of cancers of many different types are characterized by neutral evolution resulting in a characteristic power-law distribution of the mutant allele frequencies. This neutral framework provides a new way to interpret cancer genomic data and to discriminate between functional and non-functional intratumoral heterogeneity.

William Greenleaf, Michael Snyder, Carlos Araya and colleagues use density-based clustering methods on ~4,700 exomes from 21 tumor types to detect significantly mutated regions (SMRs), which show recurrent alterations in coding and noncoding elements and often associate with changes in gene expression and signaling. Mutation frequencies in SMRs demonstrate that distinct protein regions are differentially mutated across tumor types.

Yaniv Erlich and colleagues report a genome-wide survey of the contribution of short tandem repeats (STRs) to gene expression in humans and identify 2,060 significant expression STRs (eSTRs). They find that eSTRs contribute 10–15% of the cis heritability mediated by all common variants and are associated with various clinically relevant phenotypes.

Kerstin Meyer and colleagues analyze a breast cancer gene regulatory network generated using publicly available expression and ChIP-seq data sets. They identify a cluster of 36 regulons that are significantly enriched for known breast cancer risk-associated genes and propose the use of regulon activity for patient stratification.

Ludmil Alexandrov, Michael Stratton and colleagues analyze 10,250 human cancer genomes from 36 cancer types to identify mutational signatures with clock-like properties. They identify two signatures with different mutation rates that show a correlation between age at diagnosis and number of mutations in most cancer types.

Po-Ru Loh, Alkes Price and colleagues developed a fast algorithm for multicomponent, multi-trait variance-components analysis and use it to analyze the genetic architectures of schizophrenia and nine complex diseases from the PGC and GERA cohorts. Their analyses support a largely polygenic architecture for schizophrenia and significant genetic correlations for several pairs of GERA diseases.

Matthew Maurano, John Stamatoyannopoulos and colleagues identify 64,597 allelically imbalanced SNPs that influence transcription factor occupancy in vivo. Using these data, they develop a general scoring method to identify regulatory variants likely to affect transcription factor occupancy in the human genome.

Eunjung Lee, Peter Park, Dongwan Hong and colleagues report an analysis of cancer RNA sequencing data identifying approximately 900 somatic coding variants that cause disrupted splicing in cancer, leading to intron retention or exon skipping in many cases. Variants causing intron retention are enriched for loss-of-function mutations in tumor-suppressor genes.

Hilary Finucane, Brendan Bulik-Sullivan, Benjamin Neale, Alkes Price and colleagues introduce a new method, called stratified LD score regression, for partitioning heritability by functional category using genome-wide association study summary statistics. They observe a substantial enrichment of heritability in conserved regions and illustrate how this approach can provide insights into the biological basis of disease and direction for functional follow-up.

Brendan Bulik-Sullivan, Benjamin Neale, Hilary Finucane, Alkes Price and colleagues introduce a new technique for estimating genetic correlation that requires only genome-wide association summary statistics and that is not biased by sample overlap. Using this method, they find genetic correlations between anorexia nervosa and schizophrenia, and between educational attainment and autism spectrum disorder.

Christina Leslie and colleagues report an integrative analysis of the enhancer landscape and gene expression dynamics during hematopoietic differentiation. They also develop a quantitative model to predict gene expression changes from DNA sequence content and the lineage history of active enhancers and suggest a new mechanistic role for PU.1 at transition peaks during B cell specification.

Gil McVean and colleagues report a meta-analysis of Immunochip studies including over 17,000 multiple sclerosis cases and 30,000 controls, with imputation of classical HLA alleles. They find two interactions involving class II HLA alleles but no evidence for significant epistatic interactions or interactions between HLA and non-HLA risk variants.

Jian Yang and colleagues present a method, GREML-LDMS, to estimate heritability for complex human traits using whole-genome sequencing data or imputation with the 1000 Genomes Project reference panel. Using the heritability estimates from GREML-LDMS, they find that there is negligible missing heritability for human height and BMI.

Matthew Nelson and colleagues investigate how well genetic evidence for disease susceptibility predicts drug mechanisms. They find a correlation between gene products that are successful drug targets and genetic loci associated with the disease treated by the drug and predict that selecting genetically supported targets could increase the success rate of drugs in clinical development.

Michael Snyder and colleagues analyze whole-genome sequencing data from eight cancer subtypes and identify recurrent mutations in regulatory regions. They find evidence for positive selection of mutations in transcription factor binding sites near cancer-related genes.

Danielle Posthuma, Peter Visscher and colleagues report a meta-analysis of 17,804 traits based on virtually all twin studies from the last 50 years. For a majority of traits, twin resemblance seems solely due to additive genetic variation and lacks evidence for a substantial influence of shared environment or non-additive genetic variation.

Olga Troyanskaya and colleagues present genome-wide functional interaction networks for 144 human tissues and cell types. They identify important disease-gene associations by combining data from GWAS and tissue-specific networks. They also developed a webserver, GIANT, that includes multi-gene query capability, network visualization and analysis tools.

Claudio Isella and colleagues report an analysis of colorectal cancer (CRC) gene expression data from patient-derived xenografts, which they use to reconcile three commonly used CRC classification systems. They find that the stem/serrated/mesenchymal (SSM) transcriptional subtype of CRC, previously linked to poor prognosis, is driven by stromal cells rather than tumor cells.