The alpha-fetoprotein enhancer region activates the albumin and alpha-fetoprotein promoters during liver development

The four members of the albumin gene family encode the serum transport proteins albumin, α-fetoprotein, α-albumin, and vitamin D-binding protein. These genes are transcribed primarily in the liver with each having a different pattern of developmental expression. The tight linkage of these genes, particularly that of albumin, α-fetoprotein and α-albumin, and their liver-specific expression, has led to the suggestion that these genes share common regulatory elements. To directly examine whether the α-fetoprotein enhancer region could regulate the albumin gene family, expression of these genes was monitored in mice in which this region was deleted by homologous recombination. Our data indicate that this enhancer region is required for α-fetoprotein and albumin activation early in liver development and α-fetoprotein reactivation during liver regeneration, but that albumin, α-albumin, and vitamin D-binding protein expression later in hepatic development is not affected by the absence of these enhancers. We also demonstrate that RNA polymerase II loading on the α-fetoprotein and albumin promoters is reduced in the absence of this enhancer region, indicating a direct role for these enhancers in the assembly of the RNA Polymerase II complex during liver development.     

Nucleosome eviction and multiple co-factor binding predict estrogen-receptor-alpha-associated long-range interactions

Many enhancers regulate their target genes via long-distance interactions. High-throughput experiments like ChIA-PET have been developed to map such largely cell-type-specific interactions between cis-regulatory elements genome-widely. In this study, we integrated multiple types of data in order to reveal the general hidden patterns embedded in the ChIA-PET data. We found characteristic distance features related to promoter–promoter, enhancer–enhancer and insulator–insulator interactions. Although a protein may have many binding sites along the genome, our hypothesis is that those sites that share certain open chromatin structure can accommodate relatively larger protein complex consisting of specific regulatory and ‘bridging’ factors, and may be more likely to form robust long-range deoxyribonucleic acid (DNA) loops. This hypothesis was validated in the estrogen receptor alpha (ERα) ChIA-PET data. An efficient classifier was built to predict ERα-associated long-range interactions solely from the related ChIP-seq data, hence linking distal ERα-dependent enhancers to their target genes. We further applied the classifier to generate additional novel interactions, which were undetected in the original ChIA-PET paper but were validated by other independent experiments. Our work provides a new insight into the long-range chromatin interactions through deeper and integrative ChIA-PET data analysis and demonstrates DNA looping predictability from ordinary ChIP-seq data.     

ChIA-PET tool for comprehensive chromatin interaction analysis with paired-end tag sequencing

Chromatin interaction analysis with paired-end tag sequencing (ChIA-PET) is a new technology to study genome-wide long-range chromatin interactions bound by protein factors. Here we present ChIA-PET Tool, a software package for automatic processing of ChIA-PET sequence data, including linker filtering, mapping tags to reference genomes, identifying protein binding sites and chromatin interactions, and displaying the results on a graphical genome browser. ChIA-PET Tool is fast, accurate, comprehensive, user-friendly, and open source (available at http://chiapet.gis.a-star.edu.sg).     

Deep Learning for Population Genetic Inference

Given genomic variation data from multiple individuals, computing the likelihood of complex population genetic models is often infeasible. To circumvent this problem, we introduce a novel likelihood-free inference framework by applying deep learning, a powerful modern technique in machine learning. Deep learning makes use of multilayer neural networks to learn a feature-based function from the input (e.g., hundreds of correlated summary statistics of data) to the output (e.g., population genetic parameters of interest). We demonstrate that deep learning can be effectively employed for population genetic inference and learning informative features of data. As a concrete application, we focus on the challenging problem of jointly inferring natural selection and demography (in the form of a population size change history). Our method is able to separate the global nature of demography from the local nature of selection, without sequential steps for these two factors. Studying demography and selection jointly is motivated by Drosophila, where pervasive selection confounds demographic analysis. We apply our method to 197 African Drosophila melanogaster genomes from Zambia to infer both their overall demography, and regions of their genome under selection. We find many regions of the genome that have experienced hard sweeps, and fewer under selection on standing variation (soft sweep) or balancing selection. Interestingly, we find that soft sweeps and balancing selection occur more frequently closer to the centromere of each chromosome. In addition, our demographic inference suggests that previously estimated bottlenecks for African Drosophila melanogaster are too extreme.     

Shaping Skeletal Growth by Modular Regulatory Elements in the Bmp5 Gene

Cartilage and bone are formed into a remarkable range of shapes and sizes that underlie many anatomical adaptations to different lifestyles in vertebrates. Although the morphological blueprints for individual cartilage and bony structures must somehow be encoded in the genome, we currently know little about the detailed genomic mechanisms that direct precise growth patterns for particular bones. We have carried out large-scale enhancer surveys to identify the regulatory architecture controlling developmental expression of the mouse Bmp5 gene, which encodes a secreted signaling molecule required for normal morphology of specific skeletal features. Although Bmp5 is expressed in many skeletal precursors, different enhancers control expression in individual bones. Remarkably, we show here that different enhancers also exist for highly restricted spatial subdomains along the surface of individual skeletal structures, including ribs and nasal cartilages. Transgenic, null, and regulatory mutations confirm that these anatomy-specific sequences are sufficient to trigger local changes in skeletal morphology and are required for establishing normal growth rates on separate bone surfaces. Our findings suggest that individual bones are composite structures whose detailed growth patterns are built from many smaller lineage and gene expression domains. Individual enhancers in BMP genes provide a genomic mechanism for controlling precise growth domains in particular cartilages and bones, making it possible to separately regulate skeletal anatomy at highly specific locations in the body.