Technological developments have revolutionized measurements on plant genotypes and phenotypes, leading to routine production of large, complex data sets. This has led to increased efforts to extract meaning from these measurements and to integrate various data sets. Concurrently, machine learning has rapidly evolved and is now widely applied in science in general and in plant genotyping and phenotyping in particular. Here, we review the application of machine learning in the context of plant science and plant breeding. We focus on analyses at different phenotype levels, from biochemical to yield, and in connecting genotypes to these. In this way, we illustrate how machine learning offers a suite of methods that enable researchers to find meaningful patterns in relevant plant data.

Background: Genome mining for biosynthetic gene clusters (BGCs) has become an integral part of natural product discovery. The >200,000 microbial genomes now publicly available hold information on abundant novel chemistry. One way to navigate this vast genomic diversity is through comparative analysis of homologous BGCs, which allows identification of cross-species patterns that can be matched to the presence of metabolites or biological activities. However, current tools are hindered by a bottleneck caused by the expensive network-based approach used to group these BGCs into gene cluster families (GCFs). Results: Here, we introduce BiG-SLiCE, a tool designed to cluster massive numbers of BGCs. By representing them in Euclidean space, BiG-SLiCE can group BGCs into GCFs in a non-pairwise, near-linear fashion. We used BiG-SLiCE to analyze 1,225,071 BGCs collected from 209,206 publicly available microbial genomes and metagenome-assembled genomes within 10 days on a typical 36-core CPU server. We demonstrate the utility of such analyses by reconstructing a global map of secondary metabolic diversity across taxonomy to identify uncharted biosynthetic potential. BiG-SLiCE also provides a “query mode” that can efficiently place newly sequenced BGCs into previously computed GCFs, plus a powerful output visualization engine that facilitates user-friendly data exploration. Conclusions: BiG-SLiCE opens up new possibilities to accelerate natural product discovery and offers a first step towards constructing a global and searchable interconnected network of BGCs. As more genomes are sequenced from understudied taxa, more information can be mined to highlight their potentially novel chemistry. BiG-SLiCE is available via https://github.com/medema-group/bigslice.

Organ formation in animals and plants relies on precise control of cell state transitions to turn stem cell daughters into fully differentiated cells. In plants, cells cannot rearrange due to shared cell walls. Thus, differentiation progression and the accompanying cell expansion must be tightly coordinated across tissues. PLETHORA (PLT) transcription factor gradients are unique in their ability to guide the progression of cell differentiation at different positions in the growing Arabidopsis thaliana root, which contrasts with well-described transcription factor gradients in animals specifying distinct cell fates within an essentially static context. To understand the output of the PLT gradient, we studied the gene set transcriptionally controlled by PLTs. Our work reveals how the PLT gradient can regulate cell state by region-specific induction of cell proliferation genes and repression of differentiation. Moreover, PLT targets include major patterning genes and autoregulatory feedback components, enforcing their role as master regulators of organ development.

Background: Brassica rapa is an economically important crop species. During its long breeding history, a large number of morphotypes have been generated, including leafy vegetables such as Chinese cabbage and pakchoi, turnip tuber crops and oil crops. Results: To investigate the genetic variation underlying this morphological variation, we re-sequenced, assembled and annotated the genomes of two B. rapa subspecies, turnip crops (turnip) and a rapid cycling. We then analysed the two resulting genomes together with the Chinese cabbage Chiifu reference genome to obtain an impression of the B. rapa pan-genome. The number of genes with protein-coding changes between the three genotypes was lower than that among different accessions of Arabidopsis thaliana, which can be explained by the smaller effective population size of B. rapa due to its domestication. Based on orthology to a number of non-brassica species, we estimated the date of divergence among the three B. rapa morphotypes at approximately 250,000 YA, far predating Brassica domestication (5,000-10,000 YA). Conclusions: By analysing genes unique to turnip we found evidence for copy number differences in peroxidases, pointing to a role for the phenylpropanoid biosynthesis pathway in the generation of morphological variation. The estimated date of divergence among three B. rapa morphotypes implies that prior to domestication there was already considerably divergence among B. rapa genotypes. Our study thus provides two new B. rapa reference genomes, delivers a set of computer tools to analyse the resulting pan-genome and uses these to shed light on genetic drivers behind the rich morphological variation found in B. rapa.

From prokaryotes to eukaryotes, phenotypic variation, adaptation and speciation has been associated with structural variation between genomes of individuals within the same species. Many computer algorithms detecting such variations (callers) have recently been developed, spurred by the advent of the next-generation sequencing technology. Such callers mainly exploit split-read mapping or paired-end read mapping. However, as different callers are geared towards different types of structural variation, there is still no single caller that can be considered a community standard; instead, increasingly the various callers are combined in integrated pipelines. In this article, we review a wide range of callers, discuss challenges in the integration step and present a survey of pipelines used in population genomics studies. Based on our findings, we provide general recommendations on how to set-up such pipelines. Finally, we present an outlook on future challenges in structural variation detection.