Expression, genetic variation, and tissues: Human genomes show extensive genetic variation across individuals, but we have only just started documenting the effects of this variation on the regulation of gene expression. Furthermore, only a few tissues have been examined per genetic variant. In order to examine how genetic expression varies among tissues within individuals, the Genotype-Tissue Expression (GTEx) Consortium collected 1641 postmortem samples covering 54 body sites from 175 individuals. They identified quantitative genetic traits that affect gene expression and determined which of these exhibit tissue-specific expression patterns. Meléet al.measured how transcription varies among tissues, and Rivaset al.looked at how truncated protein variants affect expression across tissues.Science, this issue p.648, p.660, p.666; see also p.640

Support vector machines (SVM) are a powerful tool to analyze data with a number of predictors approximately equal or larger than the number of observations. However, originally, application of SVM to analyze biomedical data was limited because SVM was not designed to evaluate importance of predictor variables. Creating predictor models based on only the most relevant variables is essential in biomedical research. Currently, substantial work has been done to allow assessment of variable importance in SVM models but this work has focused on SVM implemented with linear kernels. The power of SVM as a prediction model is associated with the flexibility generated by use of non-linear kernels. Moreover, SVM has been extended to model survival outcomes. This paper extends the Recursive Feature Elimination (RFE) algorithm by proposing three approaches to rank variables based on non-linear SVM and SVM for survival analysis. The proposed algorithms allows visualization of each one the RFE iterations, and hence, identification of the most relevant predictors of the response variable. Using simulation studies based on time-to-event outcomes and three real datasets, we evaluate the three methods, based on pseudo-samples and kernel principal component analysis, and compare them with the original SVM-RFE algorithm for non-linear kernels. The three algorithms we proposed performed generally better than the gold standard RFE for non-linear kernels, when comparing the truly most relevant variables with the variable ranks produced by each algorithm in simulation studies. Generally, the RFE-pseudo-samples outperformed the other three methods, even when variables were assumed to be correlated in all tested scenarios. The proposed approaches can be implemented with accuracy to select variables and assess direction and strength of associations in analysis of biomedical data using SVM for categorical or time-to-event responses. Conducting variable selection and interpreting direction and strength of associations between predictors and outcomes with the proposed approaches, particularly with the RFE-pseudo-samples approach can be implemented with accuracy when analyzing biomedical data. These approaches, perform better than the classical RFE of Guyon for realistic scenarios about the structure of biomedical data.

Long noncoding RNAs (lncRNAs) are emerging as important regulators of developmental pathways. However, their roles in human cardiac precursor cell (CPC) remain unexplored. To characterize the long noncoding transcriptome during human CPC cardiac differentiation, we profiled the lncRNA transcriptome in CPCs isolated from the human fetal heart and identified 570 lncRNAs that were modulated during cardiac differentiation. Many of these were associated with active cardiac enhancer and super enhancers (SE) with their expression being correlated with proximal cardiac genes. One of the most upregulated lncRNAs was a SE-associated lncRNA that was named CARMEN, (CAR)diac (M)esoderm (E)nhancer-associated (N)oncoding RNA. CARMEN exhibits RNA-dependent enhancing activity and is upstream of the cardiac mesoderm-specifying gene regulatory network. Interestingly, CARMEN interacts with SUZ12 and EZH2, two components of the polycomb repressive complex 2 (PRC2). We demonstrate that CARMEN knockdown inhibits cardiac specification and differentiation in cardiac precursor cells independently of MIR-143 and -145 expression, two microRNAs located proximal to the enhancer sequences. Importantly, CARMEN expression was activated during pathological remodeling in the mouse and human hearts, and was necessary for maintaining cardiac identity in differentiated cardiomyocytes. This study demonstrates therefore that CARMEN is a crucial regulator of cardiac cell differentiation and homeostasis.

Genomic studies of endangered species provide insights into their evolution and demographic history, reveal patterns of genomic erosion that might limit their viability, and offer tools for their effective conservation. The Iberian lynx (Lynx pardinus) is the most endangered felid and a unique example of a species on the brink of extinction. We generate the first annotated draft of the Iberian lynx genome and carry out genome-based analyses of lynx demography, evolution, and population genetics. We identify a series of severe population bottlenecks in the history of the Iberian lynx that predate its known demographic decline during the 20th century and have greatly impacted its genome evolution. We observe drastically reduced rates of weak-to-strong substitutions associated with GC-biased gene conversion and increased rates of fixation of transposable elements. We also find multiple signatures of genetic erosion in the two remnant Iberian lynx populations, including a high frequency of potentially deleterious variants and substitutions, as well as the lowest genome-wide genetic diversity reported so far in any species. The genomic features observed in the Iberian lynx genome may hamper short- and long-term viability through reduced fitness and adaptive potential. The knowledge and resources developed in this study will boost the research on felid evolution and conservation genomics and will benefit the ongoing conservation and management of this emblematic species.

Chitinase 3-like 1 (CHI3L1) has been proposed as a biomarker associated with the conversion to clinically definite multiple sclerosis in patients with clinically isolated syndromes, based on the finding of increased cerebrospinal fluid CHI3L1 levels in clinically isolated syndrome patients who later converted to multiple sclerosis compared to those who remained as clinically isolated syndrome. Here, we aimed to validate CHI3L1 as a prognostic biomarker in a large cohort of patients with clinically isolated syndrome. This is a longitudinal cohort study of clinically isolated syndrome patients with clinical, magnetic resonance imaging, and cerebrospinal fluid data prospectively acquired. A total of 813 cerebrospinal fluid samples from patients with clinically isolated syndrome were recruited from 15 European multiple sclerosis centres. Cerebrospinal fluid CHI3L1 levels were measured by enzyme-linked immunosorbent assay. Multivariable Cox regression models were used to investigate the association between cerebrospinal fluid CHI3L1 levels and time to conversion to multiple sclerosis and time to reach Expanded Disability Status Scale 3.0. CHI3L1 levels were higher in patients who converted to clinically definite multiple sclerosis compared to patients who continued as clinically isolated syndrome (P = 8.1 × 10⁻¹¹). In the Cox regression analysis, CHI3L1 levels were a risk factor for conversion to multiple sclerosis (hazard ratio = 1.7; P = 1.1 × 10⁻⁵ using Poser criteria; hazard ratio = 1.6; P = 3.7 × 10⁻⁶ for McDonald criteria) independent of other covariates such as brain magnetic resonance imaging abnormalities and presence of cerebrospinal fluid oligoclonal bands, and were the only significant independent risk factor associated with the development of disability (hazard ratio = 3.8; P = 2.5 × 10⁻⁸). High CHI3L1 levels were associated with shorter time to multiple sclerosis (P = 3.2 × 10⁻⁹ using Poser criteria; P = 5.6 × 10⁻¹¹ for McDonald criteria) and more rapid development of disability (P = 1.8 × 10⁻¹⁰). These findings validate cerebrospinal fluid CHI3L1 as a biomarker associated with the conversion to multiple sclerosis and development of disability and reinforce the prognostic role of CHI3L1 in patients with clinically isolated syndrome. We propose that determining cerebrospinal fluid chitinase 3-like 1 levels at the time of a clinically isolated syndrome event will help identify those patients with worse disease prognosis.

Alternative splicing (AS) is a fundamental step in eukaryotic mRNA biogenesis. Here, we develop an efficient and reproducible pipeline for the discovery of genetic variants that affect AS (splicing QTLs, sQTLs). We use it to analyze the GTEx dataset, generating a comprehensive catalog of sQTLs in the human genome. Downstream analysis of this catalog provides insight into the mechanisms underlying splicing regulation. We report that a core set of sQTLs is shared across multiple tissues. sQTLs often target the global splicing pattern of genes, rather than individual splicing events. Many also affect the expression of the same or other genes, uncovering regulatory loci that act through different mechanisms. sQTLs tend to be located in post-transcriptionally spliced introns, which would function as hotspots for splicing regulation. While many variants affect splicing patterns by altering the sequence of splice sites, many more modify the binding sites of RNA-binding proteins. Genetic variants affecting splicing can have a stronger phenotypic impact than those affecting gene expression.