A popular approach for comparing gene expression levels between (replicated) conditions of RNA sequencing data relies on counting reads that map to features of interest. Within such count-based methods, many flexible and advanced statistical approaches now exist and offer the ability to adjust for covariates (e.g. batch effects). Often, these methods include some sort of ‘sharing of information’ across features to improve inferences in small samples. It is important to achieve an appropriate tradeoff between statistical power and protection against outliers. Here, we study the robustness of existing approaches for count-based differential expression analysis and propose a new strategy based on observation weights that can be used within existing frameworks. The results suggest that outliers can have a global effect on differential analyses. We demonstrate the effectiveness of our new approach with real data and simulated data that reflects properties of real datasets (e.g. dispersion-mean trend) and develop an extensible framework for comprehensive testing of current and future methods. In addition, we explore the origin of such outliers, in some cases highlighting additional biological or technical factors within the experiment. Further details can be downloaded from the project website: http://imlspenticton.uzh.ch/robinson_lab/edgeR_robust/.

Canadian Institutes of Health Research, Heart and Stroke Foundation of Canada, Ontario Ministry of Health and Long-Term Care, Ontario Ministry of Research, Innovation and Science, UK National Institute of Academic Anaesthesia, UK Clinical Research Collaboration, Australian and New Zealand College of Anaesthetists, and Monash University.

CRISPR-Cas9 enables efficient sequence-specific mutagenesis for creating somatic or germline mutants of model organisms. Key constraints in vivo remain the expression and delivery of active Cas9-sgRNA ribonucleoprotein complexes (RNPs) with minimal toxicity, variable mutagenesis efficiencies depending on targeting sequence, and high mutation mosaicism. Here, we apply in vitro assembled, fluorescent Cas9-sgRNA RNPs in solubilizing salt solution to achieve maximal mutagenesis efficiency in zebrafish embryos. MiSeq-based sequence analysis of targeted loci in individual embryos using CrispRVariants, a customized software tool for mutagenesis quantification and visualization, reveals efficient biallelic mutagenesis that reaches saturation at several tested gene loci. Such virtually complete mutagenesis exposes loss-of-function phenotypes for candidate genes in somatic mutant embryos for subsequent generation of stable germline mutants. We further show that targeting of non-coding elements in gene regulatory regions using saturating mutagenesis uncovers functional control elements in transgenic reporters and endogenous genes in injected embryos. Our results establish that optimally solubilized, in vitro assembled fluorescent Cas9-sgRNA RNPs provide a reproducible reagent for direct and scalable loss-of-function studies and applications beyond zebrafish experiments that require maximal DNA cutting efficiency in vivo.

SUMMARY Bladder cancer incurs a higher lifetime treatment cost than other cancers due to frequent recurrence of non-invasive disease. Improved prognostic biomarkers and localised therapy are needed for this large patient group. We defined two major genomic subtypes of primary stage Ta tumors that showed differential risk of recurrence. The higher risk subtype was characterised by loss of 9q including TSC1, increased KI67 labelling index, upregulated glycolysis, DNA repair, mTORC1 signaling, features of the unfolded protein response and altered cholesterol homeostasis. Comparison with muscle-invasive bladder cancer mutation profiles revealed lower overall mutation rates and more frequent mutations in RHOB and chromatin modifier genes. More mutations in the histone-lysine demethylase KDM6A were present in non-invasive tumors from females than males.

CRISPR-Cas-based genome editing holds great promise for targeting genetic disorders, including inborn errors of hepatocyte metabolism. Precise correction of disease-causing mutations in adult tissues in vivo, however, is challenging. It requires repair of Cas9-induced double-stranded DNA (dsDNA) breaks by homology-directed mechanisms, which are highly inefficient in nondividing cells. Here we corrected the disease phenotype of adult phenylalanine hydroxylase (Pah) mice, a model for the human autosomal recessive liver disease phenylketonuria (PKU), using recently developed CRISPR-Cas-associated base editors. These systems enable conversion of CG to TA base pairs and vice versa, independent of dsDNA break formation and homology-directed repair (HDR). We engineered and validated an inteinsplit base editor, which allows splitting of the fusion protein into two parts, thereby circumventing the limited cargo capacity of adeno-associated virus (AAV) vectors. Intravenous injection of AAV-base editor systems resulted in Pah gene correction rates that restored physiological blood phenylalanine (L-Phe) levels below 120 µmol/l [5]. We observed mRNA correction rates up to 63%, restoration of phenylalanine hydroxylase (PAH) enzyme activity, and reversion of the light fur phenotype in Pah mice. Our findings suggest that targeting genetic diseases in vivo using AAV-mediated delivery of base-editing agents is feasible, demonstrating potential for therapeutic application.

We have implemented in Python the COmparative GENomic Toolkit, a fully integrated and thoroughly tested framework for novel probabilistic analyses of biological sequences, devising workflows, and generating publication quality graphics. PyCogent includes connectors to remote databases, built-in generalized probabilistic techniques for working with biological sequences, and controllers for third-party applications. The toolkit takes advantage of parallel architectures and runs on a range of hardware and operating systems, and is available under the general public license from http://sourceforge.net/projects/pycogent. RationaleThe genetic divergence of species is affected by both DNA metabolic processes and natural selection. Processes contributing to genetic variation that are undetectable with intraspecific data may be detectable by inter-specific analyses because of the accumulation of signal over evolutionary time scales. As a consequence of the greater statistical power, there is interest in applying comparative analyses to address an increasing number and diversity of problems, in particular analyses that integrate sequence and phenotype. Significant barriers that hinder the extension of comparative analyses to exploit genome indexed phenotypic data include the narrow focus of most analytical tools, and the diverse array of data sources, formats, and tools available. Theoretically coherent integrative analyses can be conducted by combining probabilistic models of different aspects of genotype. Probabilistic models of sequence change underlie many core bioinformatics tasks, including similarity search, sequence alignment, phylogenetic inference, and ancestral state reconstruction. Probabilistic models allow usage of likelihood inference, a powerful approach from statistics, to establish the significance of differences in support of competing hypotheses. Linking different analyses through a shared and explicit probabilistic model of sequence change is thus extremely valuable, and provides a basis for generalizing analyses to more complex models of evolution (for example, to incorporate dependence between sites). Numerous studies have established how biological factors representing metabolic or selective influences can be represented in substitution models as specific parameters that affect rates of interchange between sequence motifs or the spatial occurrence of such rates [1][2][3][4]. Given this solid grounding, it is desirable to have a toolkit that allows flexible parameterization of probabilistic models and interchange of appropriate modules.There are many existing software packages that can manipulate biological sequences and structures, but few allow specification of both truly novel statistical models and detailed workflow control for genome scale datasets. Traditional phylogenetic analysis applications [5,6] typically provide a number of explicitly defined statistical models that are difficult to modify. One exception in which the parameterization of entirely novel substitution models was poss...

CRISPR-Cas9 and related technologies can efficiently alter genomic DNA at targeted positions and have far-reaching implications for functional screening and therapeutic gene editing. Understanding and unlocking this potential requires accurate evaluation of editing efficiency. Here, we show that methodological decisions for analyzing sequencing data can significantly affect mutagenesis efficiency estimates and we provide a comprehensive R-based toolkit, CrispRVariants and an accompanying web tool CrispRVariantsLite, that resolve and localize individual mutant alleles with respect to the endonuclease cut site. CrispRVariantsenabled analyses of newly generated and existing genome editing datasets underscore how careful consideration of the full variant spectrum gives insight toward effective guide and amplicon design as well as the mutagenic process.

Transcription of the several hundred of mouse and human Ribosomal RNA (rRNA) genes accounts for the majority of RNA synthesis in the cell nucleus and is the determinant of cytoplasmic ribosome abundance, a key factor in regulating gene expression. The rRNA genes, referred to globally as the rDNA, are clustered as direct repeats at the Nucleolar Organiser Regions, NORs, of several chromosomes, and in many cells the active repeats are transcribed at near saturation levels. The rDNA is also a hotspot of recombination and chromosome breakage, and hence understanding its control has broad importance. Despite the need for a high level of rDNA transcription, typically only a fraction of the rDNA is transcriptionally active, and some NORs are permanently silenced by CpG methylation. Various chromatin-remodelling complexes have been implicated in counteracting silencing to maintain rDNA activity. However, the chromatin structure of the active rDNA fraction is still far from clear. Here we have combined a high-resolution ChIP-Seq protocol with conditional inactivation of key basal factors to better understand what determines active rDNA chromatin. The data resolve questions concerning the interdependence of the basal transcription factors, show that preinitiation complex formation is driven by the architectural factor UBF (UBTF) independently of transcription, and that RPI termination and release corresponds with the site of TTF1 binding. They further reveal the existence of an asymmetric Enhancer Boundary Complex formed by CTCF and Cohesin and flanked upstream by phased nucleosomes and downstream by an arrested RNA Polymerase I complex. We find that the Enhancer Boundary Complex is the only site of active histone modification in the 45kbp rDNA repeat. Strikingly, it not only delimits each functional rRNA gene, but also is stably maintained after gene inactivation and the re-establishment of surrounding repressive chromatin. Our data define a poised state of rDNA chromatin and place the Enhancer Boundary Complex as the likely entry point for chromatin remodelling complexes.