RNA splicing analysis using heterogeneous and large RNA-seq datasets.

The ubiquity of RNA-seq has led to many methods that use RNA-seq data to analyze variations in RNA splicing. However, available methods are not well suited for handling heterogeneous and large datasets. Such datasets scale to thousands of samples across dozens of experimental conditions, exhibit increased variability compared to biological replicates, and involve thousands of unannotated splice variants resulting in increased transcriptome complexity. We describe here a suite of algorithms and tools implemented in the MAJIQ v2 package to address challenges in detection, quantification, and visualization of splicing variations from such datasets. Using both large scale synthetic data and GTEx v8 as benchmark datasets, we assess the advantages of MAJIQ v2 compared to existing methods. We then apply MAJIQ v2 package to analyze differential splicing across 2,335 samples from 13 brain subregions, demonstrating its ability to offer insights into brain subregion-specific splicing regulation.

Analysis of histone variant constraint and tissue expression suggests five potential novel human disease genes: H2AFY2, H2AFZ, H2AFY, H2AFV, H1F0.

While germline variants in histone protein-encoding genes are emerging as the pathogenic mutations underlying rare, Mendelian disorders characterized by a conserved phenotype of neurodevelopmental syndrome coupled with craniofacial abnormalities, a systematic assessment of all human genes encoding histone proteins has not been performed to predict novel disease-candidate genes. We first defined a comprehensive list of 89 histone-encoding genes. We then analyzed which are most likely to underlay this conserved phenotype when mutated based on their intolerance to either missense or loss-of-function variation and based on their tissue expression profile. Strikingly few genes were found to be both ubiquitously expressed and significantly constrained against missense (7.9%, n = 7) or loss-of-function (6.7%, n = 6) variation. Notably, most of those significantly constrained genes encode replication-independent, variant histone proteins (7/7 in the missense analysis, 5/6 in the loss-of-function analysis). Of the seven genes predicted to be disease-causing when germline missense variation is present, three (H2AFV, H2AFY, H2AFY2) are novel disease-candidate genes. Five of the six genes predicted to be disease-causing with an underlying germline loss-of-function variant are novel disease-candidate genes (H2AFY2, H2AFZ, H2AFY, H2AFV, H1F0). These findings may serve as a focused reference for future sequencing of patients with the conserved phenotype.

Enhanced Integrated Gradients: improving interpretability of deep learning models using splicing codes as a case study.

Despite the success and fast adaptation of deep learning models in biomedical domains, their lack of interpretability remains an issue. Here, we introduce Enhanced Integrated Gradients (EIG), a method to identify significant features associated with a specific prediction task. Using RNA splicing prediction as well as digit classification as case studies, we demonstrate that EIG improves upon the original Integrated Gradients method and produces sets of informative features. We then apply EIG to identify A1CF as a key regulator of liver-specific alternative splicing, supporting this finding with subsequent analysis of relevant A1CF functional (RNA-seq) and binding data (PAR-CLIP).

Mapping RNA splicing variations in clinically accessible and nonaccessible tissues to facilitate Mendelian disease diagnosis using RNA-seq.

PURPOSE: RNA-seq is a promising approach to improve diagnoses by detecting pathogenic aberrations in RNA splicing that are missed by DNA sequencing. RNA-seq is typically performed on clinically accessible tissues (CATs) from blood and skin. RNA tissue specificity makes it difficult to identify aberrations in relevant but nonaccessible tissues (non-CATs). We determined how RNA-seq from CATs represent splicing in and across genes and non-CATs. METHODS: We quantified RNA splicing in 801 RNA-seq samples from 56 different adult and fetal tissues from Genotype-Tissue Expression Project (GTEx) and ArrayExpress. We identified genes and splicing events in each non-CAT and determined when RNA-seq in each CAT would inadequately represent them. We developed an online resource, MAJIQ-CAT, for exploring our analysis for specific genes and tissues. RESULTS: In non-CATs, 40.2% of genes have splicing that is inadequately represented by at least one CAT; 6.3% of genes have splicing inadequately represented by all CATs. A majority (52.1%) of inadequately represented genes are lowly expressed in CATs (transcripts per million (TPM) < 1), but 5.8% are inadequately represented despite being well expressed (TPM > 10). CONCLUSION: Many splicing events in non-CATs are inadequately evaluated using RNA-seq from CATs. MAJIQ-CAT allows users to explore which accessible tissues, if any, best represent splicing in genes and tissues of interest.

Random Forest Analysis of Untargeted Metabolomics Data Suggests Increased Use of Omega Fatty Acid Oxidation Pathway in Drosophila Melanogaster Larvae Fed a Medium Chain Fatty Acid Rich High-Fat Diet.

Obesity is a complex disease, shaped by both genetic and environmental factors such as diet. In this study, we use untargeted metabolomics and Drosophila melanogaster to model how diet and genotype shape the metabolome of obese phenotypes. We used 16 distinct outbred genotypes of Drosophila larvae raised on normal (ND) and high-fat (HFD) diets, to produce three distinct phenotypic classes; genotypes that stored more triglycerides on a ND relative to the HFD, genotypes that stored more triglycerides on a HFD relative to ND, and genotypes that showed no change in triglyceride storage on either of the two diets. Using untargeted metabolomics we characterized 350 metabolites: 270 with definitive chemical IDs and 80 that were chemically unidentified. Using random forests, we determined metabolites that were important in discriminating between the HFD and ND larvae as well as between the triglyceride phenotypic classes. We found that flies fed on a HFD showed evidence of an increased use of omega fatty acid oxidation pathway, an alternative to the more commonly used beta fatty acid oxidation pathway. Additionally, we observed no correlation between the triglyceride storage phenotype and free fatty acid levels (laurate, caprate, caprylate, caproate), indicating that the distinct metabolic profile of fatty acids in high-fat diet fed Drosophila larvae does not propagate into triglyceride storage differences. However, dipeptides did show moderate differences between the phenotypic classes. We fit Gaussian graphical models (GGMs) of the metabolic profiles for HFD and ND flies to characterize changes in metabolic network structure between the two diets, finding the HFD to have a greater number of edges indicating that metabolome varies more across samples on a HFD. Taken together, these results show that, in the context of obesity, metabolomic profiles under distinct dietary conditions may not be reliable predictors of phenotypic outcomes in a genetically diverse population.

ChromoShake: a chromosome dynamics simulator reveals that chromatin loops stiffen centromeric chromatin.

ChromoShake is a three-dimensional simulator designed to find the thermodynamically favored states for given chromosome geometries. The simulator has been applied to a geometric model based on experimentally determined positions and fluctuations of DNA and the distribution of cohesin and condensin in the budding yeast centromere. Simulations of chromatin in differing initial configurations reveal novel principles for understanding the structure and function of a eukaryotic centromere. The entropic position of DNA loops mirrors their experimental position, consistent with their radial displacement from the spindle axis. The barrel-like distribution of cohesin complexes surrounding the central spindle in metaphase is a consequence of the size of the DNA loops within the pericentromere to which cohesin is bound. Linkage between DNA loops of different centromeres is requisite to recapitulate experimentally determined correlations in DNA motion. The consequences of radial loops and cohesin and condensin binding are to stiffen the DNA along the spindle axis, imparting an active function to the centromere in mitosis.

Variant non ketotic hyperglycinemia is caused by mutations in LIAS, BOLA3 and the novel gene GLRX5.

Patients with nonketotic hyperglycinemia and deficient glycine cleavage enzyme activity, but without mutations in AMT, GLDC or GCSH, the genes encoding its constituent proteins, constitute a clinical group which we call 'variant nonketotic hyperglycinemia'. We hypothesize that in some patients the aetiology involves genetic mutations that result in a deficiency of the cofactor lipoate, and sequenced genes involved in lipoate synthesis and iron-sulphur cluster biogenesis. Of 11 individuals identified with variant nonketotic hyperglycinemia, we were able to determine the genetic aetiology in eight patients and delineate the clinical and biochemical phenotypes. Mutations were identified in the genes for lipoate synthase (LIAS), BolA type 3 (BOLA3), and a novel gene glutaredoxin 5 (GLRX5). Patients with GLRX5-associated variant nonketotic hyperglycinemia had normal development with childhood-onset spastic paraplegia, spinal lesion, and optic atrophy. Clinical features of BOLA3-associated variant nonketotic hyperglycinemia include severe neurodegeneration after a period of normal development. Additional features include leukodystrophy, cardiomyopathy and optic atrophy. Patients with lipoate synthase-deficient variant nonketotic hyperglycinemia varied in severity from mild static encephalopathy to Leigh disease and cortical involvement. All patients had high serum and borderline elevated cerebrospinal fluid glycine and cerebrospinal fluid:plasma glycine ratio, and deficient glycine cleavage enzyme activity. They had low pyruvate dehydrogenase enzyme activity but most did not have lactic acidosis. Patients were deficient in lipoylation of mitochondrial proteins. There were minimal and inconsistent changes in cellular iron handling, and respiratory chain activity was unaffected. Identified mutations were phylogenetically conserved, and transfection with native genes corrected the biochemical deficiency proving pathogenicity. Treatments of cells with lipoate and with mitochondrially-targeted lipoate were unsuccessful at correcting the deficiency. The recognition of variant nonketotic hyperglycinemia is important for physicians evaluating patients with abnormalities in glycine as this will affect the genetic causation and genetic counselling, and provide prognostic information on the expected phenotypic course.