Modeling chromatin state from sequence across angiosperms using recurrent convolutional neural networks

Accessible chromatin regions are critical components of gene regulation but modeling them directly from sequence remains challenging, especially within plants, whose mechanisms of chromatin remodeling are less understood than in animals. We trained an existing deep learning architecture, DanQ, on leaf ATAC-seq data from 12 angiosperm species to predict the chromatin accessibility of sequence windows within and across species. We also trained DanQ on DNA methylation data from 10 angiosperms, because unmethylated regions have been shown to overlap significantly with accessible chromatin regions in some plants. The across-species models have comparable or even superior performance to a model trained within species, suggesting strong conservation of chromatin mechanisms across angiosperms. Testing a maize held out model on a multi-tissue scATAC panel revealed our models are best at predicting constitutively-accessible chromatin regions, with diminishing performance as cell-type specificity increases. Using a combination of interpretation methods, we ranked JASPAR motifs by their importance to each model and saw that the TCP and AP2/ERF transcription factor families consistently ranked highly. We embedded the top three JASPAR motifs for each model at all possible positions on both strands in our sequence window and observed position- and strand-specific patterns in their importance to the model. With our cross-species "a2z" model it is now feasible to predict the chromatin accessibility and methylation landscape of any angiosperm genome.

Insights into angiosperm evolution, floral development and chemical biosynthesis from the Aristolochia fimbriata genome

Aristolochia, a genus in the magnoliid order Piperales, has been famous for centuries for its highly specialized flowers and wide medicinal applications. Here, we present a new, high-quality genome sequence of Aristolochia fimbriata, a species that, similar to Amborella trichopoda, lacks further whole-genome duplications since the origin of extant angiosperms. As such, the A. fimbriata genome is an excellent reference for inferences of angiosperm genome evolution, enabling detection of two novel whole-genome duplications in Piperales and dating of previously reported whole-genome duplications in other magnoliids. Genomic comparisons between A. fimbriata and other angiosperms facilitated the identification of ancient genomic rearrangements suggesting the placement of magnoliids as sister to monocots, whereas phylogenetic inferences based on sequence data we compiled yielded ambiguous relationships. By identifying associated homologues and investigating their evolutionary histories and expression patterns, we revealed highly conserved floral developmental genes and their distinct downstream regulatory network that may contribute to the complex flower morphology in A. fimbriata. Finally, we elucidated the genetic basis underlying the biosynthesis of terpenoids and aristolochic acids in A. fimbriata.

Comparison of gene order of GIGANTEA loci in yellow-poplar, monocots, and eudicots

GIGANTEA plays an important role in the control of circadian rhythms and photoperiodic flowering. The GIGANTEA gene has been studied in various species, but not in basal angiosperms. Moreover, to the best of our knowledge, no study of the genome organization of a basal angiosperm has yet been published. In this study, we sequenced a bacterial artificial chromosome (BAC) harboring GIGANTEA from yellow-poplar ( Liriodendron tulipifera L.) and compared the genomic organization of this gene in yellow-poplar with that in other species from various angiosperm clades. This is the first report on the gene structure and organization of a large contig in any basal angiosperm species. The BAC clone, covering a region of approximately 122 kb from the yellow-poplar genome, was sequenced and assembled by coupling the 454 pyrosequencing technology with ABI capillary sequencing. In addition to GIGANTEA, the gene RPS18.A (encoding ribosomal protein S18.A) was found in this segment of the genome. We found that gene content and order in this region of the yellow-poplar genome were similar to those in the corresponding region in eudicots but not in Oryza sativa and Sorghum bicolor , implying that clustering of the GIGANTEA and RPS18.A genes is ancestral and separation of the genes occurred after the phylogenetic split of monocots from dicots. Phylogenetic analysis of GIGANTEA amino acid sequences placed yellow-poplar closer to eudicots than to monocots. In addition, evidence for transposition and large insertions and duplications was found, suggesting multiple and complex mechanisms of basal angiosperm genome evolution.