Search

In the past two decades genome sequencing has developed from a laborious and costly technology employed by large international consortia to a widely used, automated and affordable tool used worldwide by many individual research groups. Genome sequences of many food animals and crop plants have been deciphered and are being exploited for fundamental research and applied to improve their breeding programs. The developments in sequencing technologies have also impacted the associated bioinformatics strategies and tools, both those that are required for data processing, management, and quality control, and those used for interpretation of the data. This thesis focuses on the application of genome sequencing, assembly and annotation to two members of the Solanaceae family, tomato and potato. Potato is the economically most important species within the Solanaceae, and its tubers contribute to dietary intake of starch, protein, antioxidants, and vitamins. Tomato fruits are the second most consumed vegetable after potato, and are a globally important dietary source of lycopene, beta-carotene, vitamin C, and fiber. The chapters in this thesis document the generation, exploitation and interpretation of genomic sequence resources for these two species and shed light on the contents, structure and evolution of their genomes. Chapter 1introduces the concepts of genome sequencing, assembly and annotation, and explains the novel genome sequencing technologies that have been developed in the past decade. These so-called Next Generation Sequencing platforms display considerable variation in chemistry and workflow, and as a consequence the throughput and data quality differs by orders of magnitude between the platforms. The currently available sequencing platforms produce a vast variety of read lengths and facilitate the generation of paired sequences with an approximately fixed distance between them. The choice of sequencing chemistry and platform combined with the type of sequencing template demands specifically adapted bioinformatics for data processing and interpretation. Irrespective of the sequencing and assembly strategy that is chosen, the resulting genome sequence, often represented by a collection of long linear strings of nucleotides, is of limited interest by itself. Interpretation of the genome can only be achieved through sequence annotation – that is, identification and classification of all functional elements in a genome sequence. Once these elements have been annotated, sequence alignments between multiple genomes of related accessions or species can be utilized to reveal the genetic variation on both the nucleotide and the structural level that underlies the difference between these species or accessions. Chapter 2describes BlastIf, a novel software tool that exploits sequence similarity searches with BLAST to provide a straightforward annotation of long nucleotide sequences. Generally, two problems are associated with the alignment of a long nucleotide sequence to a database of short gene or protein sequences: (i) the large number of similar hits that can be generated due to database redundancy; and (ii) the relationships implied between aligned segments within a hit that in fact correspond to distinct elements on the sequence such as genes. BlastIf generates a comprehensible BLAST output for long nucleotide sequences by reducing the number of similar hits while revealing most of the variation present between hits. It is a valuable tool for molecular biologists who wish to get a quick overview of the genetic elements present in a newly sequenced segment of DNA, prior to more elaborate efforts of gene structure prediction and annotation. In Chapter 3 a first genome-wide comparison between the emerging genomic sequence resources of tomato and potato is presented. Large collections of BAC end sequences from both species were annotated through repeat searches, transcript alignments and protein domain identification. In-depth comparisons of the annotated sequences revealed remarkable differences in both gene and repeat content between these closely related genomes. The tomato genome was found to be more repetitive than the potato genome, and substantial differences in the distribution of Gypsy and Copia retrotransposable elements as well as microsatellites were observed between the two genomes. A higher gene content was identified in the potato sequences, and in particular several large gene families including cytochrome P450 mono-oxygenases and serine-threonine protein kinases were significantly overrepresented in potato compared to tomato. Moreover, the cytochrome P450 gene family was found to be expanded in both tomato and potato when compared to Arabidopsis thaliana, suggesting an expanded network of secondary metabolic pathways in the Solanaceae. Together these findings present a first glimpse into the evolution of Solanaceous genomes, both within the family and relative to other plant species. Chapter 4explores the physical and genetic organization of tomato chromosome 6 through integration of BAC sequence analysis, High Information Content Fingerprinting, genetic analysis, and BAC-FISH mapping data. A collection of BACs spanning substantial parts of the short and long arm euchromatin and several dispersed regions of the pericentrometric heterochromatin were sequenced and assembled into several tiling paths spanning approximately 11 Mb. Overall, the cytogenetic order of BACs was in agreement with the order of BACs anchored to the Tomato EXPEN 2000 genetic map, although a few striking discrepancies were observed. The integration of BAC-FISH, sequence and genetic mapping data furthermore provided a clear picture of the borders between eu- and heterochromatin on chromosome 6. Annotation of the BAC sequences revealed that, although the majority of protein-coding genes were located in the euchromatin, the highly repetitive pericentromeric heterochromatin displayed an unexpectedly high gene content. Moreover, the short arm euchromatin was relatively rich in repeats, but the ratio of Gypsy and Copia retrotransposons across the different domains of the chromosome clearly distinguished euchromatin from heterochromatin. The ongoing whole-genome sequencing effort will reveal if these properties are unique for tomato chromosome 6, or a more general property of the tomato genome. Chapter 5presents the potato genome, the first genome sequence of an Asterid. To overcome the problems associated with genome assembly due tothe high level of heterozygosity that is observed in commercial tetraploid potato varieties, a homozygous doubled-monoploid potato clone was exploited to sequence and assemble 86% of the 844 Mb genome. This potato reference genome sequence was complemented with re-sequencing of aheterozygous diploid clone, revealing the form and extent of sequence polymorphism both between different genotypes and within a single heterozygous genotype. Gene presence/absence variants and other potentially deleterious mutations were found to occur frequently in potato and are a likely cause of inbreeding depression. Annotation of the genome was supported by deep transcriptome sequencing of both the doubled-monoploid and the heterozygous potato, resulting in the prediction of more than 39,000 protein coding genes. Transcriptome analysis provided evidence for the contribution of gene family expansion, tissue specific expression, and recruitment of genes to new pathways to the evolution of tuber development. The sequence of the potato genome has provided new insights into Eudicot genome evolution and has provided a solid basis for the elucidation of the evolution of tuberisation. Many traits of interest to plant breeders are quantitative in nature and the potato sequence will simplify both their characterization and deployment to generate novel cultivars. The outstanding challenges in plant genome sequencing are addressed in Chapter 6. The high concentration of repetitive elements and the heterozygosity and polyploidy of many interesting crop plant species currently pose a barrier for the efficient reconstruction of their genome sequences. Nonetheless, the completion of a large number of new genome sequences in recent years and the ongoing advances in sequencing technology provide many excitingopportunities for plant breeding and genome research. Current sequencing platforms are being continuously updated and improved, and novel technologies are being developed and implemented in third-generation sequencing platforms that sequence individual molecules without need for amplification. While these technologies create exciting opportunities for new sequencing applications, they also require robust software tools to process the data produced through them efficiently. The ever increasing amount of available genome sequences creates the need for an intuitive platform for the automated and reproducible interrogation of these data in order to formulate new biologically relevant questions on datasets spanning hundreds or thousands of genome sequences.

Splicing is one of the key processing steps during the maturation of a gene’s primary transcript into the mRNA molecule used as a template for protein production. Splicing involves the removal of segments called introns and re-joining of the remaining segments called exons. It is by now well established that not always the same segments are removed from a gene’s primary transcript during the splicing process. The consequence of this splicing variation, termed Alternative Splicing (AS), is that multiple distinct mature mRNA molecules can be produced from a single gene. One of the two biological roles that are ascribed to AS is that of a mechanism which enables an organism to produce multiple functionally distinct proteins from a single gene. Alternatively, AS can serve as a means for controlling gene expression at the post-transcriptional level. Although many clear examples have been reported for both roles, the extent to which AS increases the functional diversity of the proteome, regulates gene expression or simply reflects noise in splicing machinery is not well known. Determining the full functional impact of AS by designing and performing wet-lab experiments for all AS events is unfeasible and bioinformatics approaches have therefore widely been used for studying the impact of AS at a genome-wide scale. In this thesis four bioinformatics studies are presented that were aimed at determining the extent to which AS is used in plants as a mechanism for producing multiple distinct functional proteins from a single gene. Each chapter uses a different method for analyzing specific properties of AS. Under the premise that functional genetic features are more likely to be conserved than non-functional ones, AS events that are present in two or more species are more likely to be biologically relevant than those that are confined to a single species. In chapter 2 we analyzed the conservation of AS by performing a comparative analysis between three divergent plant species. The results of that study indicated that the vast majority of AS events does not persist over long periods of evolution. We concluded, based on this lack of conservation, that AS only has a limited impact on the functional diversity of the proteome in plants. Following this conclusion, it can hypothesized that the variation that AS induces at the transcriptome level is not likely to be manifested at the protein level. In chapter 3 we tested this hypothesis by analyzing two independent proteomics datasets. This type of data can be used to directly identify proteins present in a biological sample. Our results indicated that the variation induced by AS at the transcriptome level is also manifested at the protein level. We concluded that either many AS events have a confined species-specific (not conserved) function or simply produce protein variants that are stable enough to escape rapid turn-over. Another method for determining whether AS increases the functional diversity of the proteome is by determining whether protein sequence variations that are typically induced by AS are common within the plant kingdom. We found (chapter 4) that this is not the case in plants and concluded that novel functions do not frequently arise through AS. We also found that most of the AS-induced variation is lost, similarly as for redundant gene copies, within a very short evolutionary time period. One limitation of genome-wide analyses is that these capture only the more general patterns. However, the functional impact of AS can be very different in different genes or gene-families. In order fully assess the functional impact of AS, it is therefore important to also study the process within the functional context of individual genes or gene families. In chapter 5 we demonstrated this concept by performing a detailed analysis of AS within the MADS-box gene family. We were able to provide clues as to how AS might impact the protein-protein interaction capabilities of individual MADS proteins. Some of our predictions were supported by experimental evidence. We further showed how AS can serve as an evolutionary mechanism for experimenting with novel functions (novel interactions) without the explicit loss of existing functions. The overall conclusion, based on the performed analyses is as follows: AS primarily is a consequence of noise in the splicing machinery and results in an increased diversity of the proteome. However, only a small fraction of the proteins resulting from AS will have beneficial functions and are subsequently selected for during evolution. The large remaining fraction is, similarly as for redundant gene-copies, lost within a very short evolutionary time period after its emergence.

Unravelling the functions of proteins is one of the most important aims of modern biology. Experimental inference of protein function is expensive and not scalable to large datasets. In this thesis a probabilistic method for protein function prediction is presented that integrates different types of data such as sequences and networks. The method is based on Bayesian Markov Random Field (BMRF) analysis. BMRF was initially applied to genome wide protein function prediction using network data in yeast and in also in Arabidopsis by integrating protein domains (i.e InterPro signatures), expressions and protein protein interactions. Several of the predictions were confirmed by experimental evidence. Further, an evolutionary discrete optimization algorithm is presented that integrates function predictions from different Gene Ontology (GO) terms to a single prediction that is consistent to the True Path Rule as imposed by the GO Directed Acyclic Graph. This integration leads to predictions that are easy to be interpreted. Evaluation of of this algorithm using Arabidopsis data showed that the prediction performance is improved, compared to single GO term predictions.

The quest for understanding how proteins evolve and function has been a prominent and costly human endeavor. With advances in genomics and use of bioinformatics tools, the diversity of proteins in present day genomes can now be studied more efficiently than ever before. This thesis describes computational methods suitable for large-scale protein classification of many proteomes of diverse species. Specifically, we focus on methods that combine unsupervised learning (clustering) techniques with the knowledge of molecular phylogenetics, particularly that of orthology. In chapter 1 we introduce the biological context of protein structure, function and evolution, review the state-of-the-art sequence-based protein classification methods, and then describe methods used to validate the predictions. Finally, we present the outline and objectives of this thesis. Evolutionary (phylogenetic) concepts are instrumental in studying subjects as diverse as the diversity of genomes, cellular networks, protein structures and functions, and functional genome annotation. In particular, the detection of orthologous proteins (genes) across genomes provides reliable means to infer biological functions and processes from one organism to another. Chapter 2 evaluates the available computational tools, such as algorithms and databases, used to infer orthologous relationships between genes from fully sequenced genomes. We discuss the main caveats of large-scale orthology detection in general as well as the merits and pitfalls of each method in particular. We argue that establishing true orthologous relationships requires a phylogenetic approach which combines both trees and graphs (networks), reliable species phylogeny, genomic data for more than two species, and an insight into the processes of molecular evolution. Also proposed is a set of guidelines to aid researchers in selecting the correct tool. Moreover, this review motivates further research in developing reliable and scalable methods for functional and phylogenetic classification of large protein collections. Chapter 3 proposes a framework in which various protein knowledge-bases are combined into unique network of mappings (links), and hence allows comparisons to be made between expert curated and fully-automated protein classifications from a single entry point. We developed an integrated annotation resource for protein orthology, ProGMap (Protein Group Mappings, http://www.bioinformatics.nl/progmap), to help researchers and database annotators who often need to assess the coherence of proposed annotations and/or group assignments, as well as users of high throughput methodologies (e.g., microarrays or proteomics) who deal with partially annotated genomic data. ProGMap is based on a non-redundant dataset of over 6.6 million protein sequences which is mapped to 240,000 protein group descriptions collected from UniProt, RefSeq, Ensembl, COG, KOG, OrthoMCL-DB, HomoloGene, TRIBES and PIRSF using a fast and fully automated sequence-based mapping approach. The ProGMap database is equipped with a web interface that enables queries to be made using synonymous sequence identifiers, gene symbols, protein functions, and amino acid or nucleotide sequences. It incorporates also services, namely BLAST similarity search and QuickMatch identity search, for finding sequences similar (or identical) to a query sequence, and tools for presenting the results in graphic form. Graphs (networks) have gained an increasing attention in contemporary biology because they have enabled complex biological systems and processes to be modeled and better understood. For example, protein similarity networks constructed of all-versus-all sequence comparisons are frequently used to delineate similarity groups, such as protein families or orthologous groups in comparative genomics studies. Chapter 4.1 presents a benchmark study of freely available graph software used for this purpose. Specifically, the computational complexity of the programs is investigated using both simulated and biological networks. We show that most available software is not suitable for large networks, such as those encountered in large-scale proteome analyzes, because of the high demands on computational resources. To address this, we developed a fast and memory-efficient graph software, netclust (http://www.bioinformatics.nl/netclust/), which can scale to large protein networks, such as those constructed of millions of proteins and sequence similarities, on a standard computer. An extended version of this program called Multi-netclust is presented in chapter 4.2. This tool that can find connected clusters of data presented by different network data sets. It uses user-defined threshold values to combine the data sets in such a way that clusters connected in all or in either of the networks can be retrieved efficiently. Automated protein sequence clustering is an important task in genome annotation projects and phylogenomic studies. During the past years, several protein clustering programs have been developed for delineating protein families or orthologous groups from large sequence collections. However, most of these programs have not been benchmarked systematically, in particular with respect to the trade-off between computational complexity and biological soundness. In chapter 5 we evaluate three best known algorithms on different protein similarity networks and validation (or 'gold' standard) data sets to find out which one can scale to hundreds of proteomes and still delineate high quality similarity groups at the minimum computational cost. For this, a reliable partition-based approach was used to assess the biological soundness of predicted groups using known protein functions, manually curated protein/domain families and orthologous groups available in expert-curated databases. Our benchmark results support the view that a simple and computationally cheap method such as netclust can perform similar to and in cases even better than more sophisticated, yet much more costly methods. Moreover, we introduce an efficient graph-based method that can delineate protein orthologs of hundreds of proteomes into hierarchical similarity groups de novo. The validity of this method is demonstrated on data obtained from 347 prokaryotic proteomes. The resulting hierarchical protein classification is not only in agreement with manually curated classifications but also provides an enriched framework in which the functional and evolutionary relationships between proteins can be studied at various levels of specificity. Finally, in chapter 6 we summarize the main findings and discuss the merits and shortcomings of the methods developed herein. We also propose directions for future research. The ever increasing flood of new sequence data makes it clear that we need improved tools to be able to handle and extract relevant (orthological) information from these protein data. This thesis summarizes these needs and how they can be addressed by the available tools, or be improved by the new tools that were developed in the course of this research.