Building a genome analysis pipeline to predict disease risk and prevent disease

funtrp: identifying protein positions for variation driven functional tuning

Evaluating the impact of non-synonymous genetic variants is essential for uncovering disease associations and mechanisms of evolution. An in-depth understanding of sequence changes is also fundamental for synthetic protein design and stability assessments. However, the variant effect predictor performance gain observed in recent years has not kept up with the increased complexity of new methods. One likely reason for this might be that most approaches use similar sets of gene and protein features for modeling variant effects, often emphasizing sequence conservation. While high levels of conservation highlight residues essential for protein activity, much of the variation observable in vivo is arguably weaker in its impact, thus requiring evaluation at a higher level of resolution. Here, we describe functionNeutral/Toggle/Rheostatpredictor (funtrp), a novel computational method that categorizes protein positions based on the position-specific expected range of mutational impacts: Neutral (weak/no effects), Rheostat (function-tuning positions), or Toggle (on/off switches). We show that position types do not correlate strongly with familiar protein features such as conservation or protein disorder. We also find that position type distribution varies across different protein functions. Finally, we demonstrate that position types can improve performance of existing variant effect predictors and suggest a way forward for the development of new ones.

Toward achieving precision health

Health care systems primarily focus on patients after they present with disease, not before. The emerging field of precision health encourages disease prevention and earlier detection by monitoring health and disease based on an individual's risk. Active participation in health care can be encouraged with continuous health-monitoring devices, providing a higher-resolution picture of human health and disease. However, the development of monitoring technologies must prioritize the collection of actionable data and long-term user engagement.

Associating mutations causing cystinuria with disease severity with the aim of providing precision medicine

Background

Cystinuria is an inherited disease that results in the formation of cystine stones in the kidney, which can have serious health complications. Two genes (SLC7A9 and SLC3A1) that form an amino acid transporter are known to be responsible for the disease. Variants that cause the disease disrupt amino acid transport across the cell membrane, leading to the build-up of relatively insoluble cystine, resulting in formation of stones. Assessing the effects of each mutation is critical in order to provide tailored treatment options for patients. We used various computational methods to assess the effects of cystinuria associated mutations, utilising information on protein function, evolutionary conservation and natural population variation of the two genes. We also analysed the ability of some methods to predict the phenotypes of individuals with cystinuria, based on their genotypes, and compared this to clinical data.

Results

Using a literature search, we collated a set of 94 SLC3A1 and 58 SLC7A9 point mutations known to be associated with cystinuria. There are differences in sequence location, evolutionary conservation, allele frequency, and predicted effect on protein function between these mutations and other genetic variants of the same genes that occur in a large population. Structural analysis considered how these mutations might lead to cystinuria. For SLC7A9, many mutations swap hydrophobic amino acids for charged amino acids or vice versa, while others affect known functional sites. For SLC3A1, functional information is currently insufficient to make confident predictions but mutations often result in the loss of hydrogen bonds and largely appear to affect protein stability. Finally, we showed that computational predictions of mutation severity were significantly correlated with the disease phenotypes of patients from a clinical study, despite different methods disagreeing for some of their predictions.

Conclusions

The results of this study are promising and highlight the areas of research which must now be pursued to better understand how mutations in SLC3A1 and SLC7A9 cause cystinuria. The application of our approach to a larger data set is essential, but we have shown that computational methods could play an important role in designing more effective personalised treatment options for patients with cystinuria.

Electronic supplementary material

The online version of this article (doi:10.1186/s12864-017-3913-1) contains supplementary material, which is available to authorized users.

Computational predictors fail to identify amino acid substitution effects at rheostat positions

Many computational approaches exist for predicting the effects of amino acid substitutions. Here, we considered whether the protein sequence position class - rheostat or toggle - affects these predictions. The classes are defined as follows: experimentally evaluated effects of amino acid substitutions at toggle positions are binary, while rheostat positions show progressive changes. For substitutions in the LacI protein, all evaluated methods failed two key expectations: toggle neutrals were incorrectly predicted as more non-neutral than rheostat non-neutrals, while toggle and rheostat neutrals were incorrectly predicted to be different. However, toggle non-neutrals were distinct from rheostat neutrals. Since many toggle positions are conserved, and most rheostats are not, predictors appear to annotate position conservation better than mutational effect. This finding can explain the well-known observation that predictors assign disproportionate weight to conservation, as well as the field's inability to improve predictor performance. Thus, building reliable predictors requires distinguishing between rheostat and toggle positions.

Protein function in precision medicine: deep understanding with machine learning

Precision medicine and personalized health efforts propose leveraging complex molecular, medical and family history, along with other types of personal data toward better life. We argue that this ambitious objective will require advanced and specialized machine learning solutions. Simply skimming some low-hanging results off the data wealth might have limited potential. Instead, we need to better understand all parts of the system to define medically relevant causes and effects: how do particular sequence variants affect particular proteins and pathways? How do these effects, in turn, cause the health or disease-related phenotype? Toward this end, deeper understanding will not simply diffuse from deeper machine learning, but from more explicit focus on understanding protein function, context-specific protein interaction networks, and impact of variation on both.

Genomic Copy Number Variation Affecting Genes Involved in the Cell Cycle Pathway: Implications for Somatic Mosaicism

Somatic genome variations (mosaicism) seem to represent a common mechanism for human intercellular/interindividual diversity in health and disease. However, origins and mechanisms of somatic mosaicism remain a matter of conjecture. Recently, it has been hypothesized that zygotic genomic variation naturally occurring in humans is likely to predispose to nonheritable genetic changes (aneuploidy) acquired during the lifetime through affecting cell cycle regulation, genome stability maintenance, and related pathways. Here, we have evaluated genomic copy number variation (CNV) in genes implicated in the cell cycle pathway (according to Kyoto Encyclopedia of Genes and Genomes/KEGG) within a cohort of patients with intellectual disability, autism, and/or epilepsy, in which the phenotype was not associated with genomic rearrangements altering this pathway. Benign CNVs affecting 20 genes of the cell cycle pathway were detected in 161 out of 255 patients (71.6%). Among them, 62 individuals exhibited >2 CNVs affecting the cell cycle pathway. Taking into account the number of individuals demonstrating CNV of these genes, a support for this hypothesis appears to be presented. Accordingly, we speculate that further studies of CNV burden across the genes implicated in related pathways might clarify whether zygotic genomic variation generates somatic mosaicism in health and disease.

Whole-Genome Sequencing and Integrative Genomic Analysis Approach on Two 22q11.2 Deletion Syndrome Family Trios for Genotype to Phenotype Correlations

The 22q11.2 deletion syndrome (22q11DS) affects 1:4,000 live births and presents with highly variable phenotype expressivity. In this study, we developed an analytical approach utilizing whole-genome sequencing (WGS) and integrative analysis to discover genetic modifiers. Our pipeline combined available tools in order to prioritize rare, predicted deleterious, coding and noncoding single-nucleotide variants (SNVs), and insertion/deletions from WGS. We sequenced two unrelated probands with 22q11DS, with contrasting clinical findings, and their unaffected parents. Proband P1 had cognitive impairment, psychotic episodes, anxiety, and tetralogy of Fallot (TOF), whereas proband P2 had juvenile rheumatoid arthritis but no other major clinical findings. In P1, we identified common variants in COMT and PRODH on 22q11.2 as well as rare potentially deleterious DNA variants in other behavioral/neurocognitive genes. We also identified a de novo SNV in ADNP2 (NM_014913.3:c.2243G>C), encoding a neuroprotective protein that may be involved in behavioral disorders. In P2, we identified a novel nonsynonymous SNV in ZFPM2 (NM_012082.3:c.1576C>T), a known causative gene for TOF, which may act as a protective variant downstream of TBX1, haploinsufficiency of which is responsible for congenital heart disease in individuals with 22q11DS.

First step in using molecular data for microbial food safety risk assessment; hazard identification of Escherichia coli O157:H7 by coupling genomic data with in vitro adherence to human epithelial cells

The potential for using whole genome sequencing (WGS) data in microbiological risk assessment (MRA) has been discussed on several occasions since the beginning of this century. Still, the proposed heuristic approaches have never been applied in a practical framework. This is due to the non-trivial problem of mapping microbial information consisting of thousands of loci onto a probabilistic scale for risks. The paradigm change for MRA involves translation of multidimensional microbial genotypic information to much reduced (integrated) phenotypic information and onwards to a single measure of human risk (i.e. probability of illness).

In this paper a first approach in methodology development is described for the application of WGS data in MRA; this is supported by a practical example. That is, combining genetic data (single nucleotide polymorphisms; SNPs) for Shiga toxin-producing Escherichia coli (STEC) O157 with phenotypic data (in vitro adherence to epithelial cells as a proxy for virulence) leads to hazard identification in a Genome Wide Association Study (GWAS).

This application revealed practical implications when using SNP data for MRA. These can be summarized by considering the following main issues: optimum sample size for valid inference on population level, correction for population structure, quantification and calibration of results, reproducibility of the analysis, links with epidemiological data, anchoring and integration of results into a systems biology approach for the translation of molecular studies to human health risk.

Future developments in genetic data analysis for MRA should aim at resolving the mapping problem of processing genetic sequences to come to a quantitative description of risk. The development of a clustering scheme focusing on biologically relevant information of the microbe involved would be a useful approach in molecular data reduction for risk assessment.

Single nucleotide variations: biological impact and theoretical interpretation

Genome-wide association studies (GWAS) and whole-exome sequencing (WES) generate massive amounts of genomic variant information, and a major challenge is to identify which variations drive disease or contribute to phenotypic traits. Because the majority of known disease-causing mutations are exonic non-synonymous single nucleotide variations (nsSNVs), most studies focus on whether these nsSNVs affect protein function. Computational studies show that the impact of nsSNVs on protein function reflects sequence homology and structural information and predict the impact through statistical methods, machine learning techniques, or models of protein evolution. Here, we review impact prediction methods and discuss their underlying principles, their advantages and limitations, and how they compare to and complement one another. Finally, we present current applications and future directions for these methods in biological research and medical genetics.

Targeted next-generation sequencing: the clinician's stethoscope for genetic disorders

Genetic biomarkers are crucial for diagnosis, guiding of treatments and estimation of prognosis. In the past, clinical genetic diagnostics was limited by the sequencing information gained from selected exons and single genes. For genetically heterogeneous diseases, such as cardiomyopathies, where underlying mutations in more than 1000 exons are known, a Sanger-based comprehensive test would have been extremely expensive and labor intensive. Next-generation sequencing has overcome these problems in terms of costs, speed and throughput. In this review we discuss available methods for targeted next-generation sequencing that ease the introduction of this technology into routine clinical application. We further provide results of a study we have performed to compare two state-of-the-art methods for their enrichment efficiency and detection accuracy of variants in a clinical setting.

In silico identification and experimental validation of insertion-deletion polymorphisms in tomato genome

Comparative analysis of the genome sequences of Solanum lycopersicum variety Heinz 1706 and S. pimpinellifolium accession LA 1589 using MUGSY software identified 145 695 insertion-deletion (InDel) polymorphisms. A selected set of 3029 candidate InDels (≥2 bp) across the entire tomato genome were subjected to PCR validation, and 82.4% could be verified. Of 2272 polymorphic InDels between LA 1589 and Heinz 1706, 61.6, 45.2, and 31.6% were polymorphic in 8 accessions of S. pimpinellifolium, 4 accessions of S. lycopersicum var. cerasiforme, and 10 varieties of S. lycopersicum, respectively. Genetic distance was 0.216 in S. pimpinellifolium, 0.202 in S. lycopersicum var. cerasiforme, and 0.108 in S. lycopersicum. The data suggested a reduction of genetic variation from S. pimpinellifolium to S. lycopersicum var. cerasiforme and S. lycopersicum. Cluster analysis showed that the 8 accessions of S. pimpinellifolium were in one group, whereas 4 accessions of S. lycopersicum var. cerasiforme and 10 varieties of S. lycopersicum were in the same group.

Advances in Human Biology: Combining Genetics and Molecular Biophysics to Pave the Way for Personalized Diagnostics and Medicine

Advances in several biology-oriented initiatives such as genome sequencing and structural genomics, along with the progress made through traditional biological and biochemical research, have opened up a unique opportunity to better understand the molecular effects of human diseases. Human DNA can vary significantly from person to person and determines an individual’s physical characteristics and their susceptibility to diseases. Armed with an individual’s DNA sequence, researchers and physicians can check for defects known to be associated with certain diseases by utilizing various databases. However, for unclassified DNA mutations or in order to reveal molecular mechanism behind the effects, the mutations have to be mapped onto the corresponding networks and macromolecular structures and then analyzed to reveal their effect on the wild type properties of biological processes involved. Predicting the effect of DNA mutations on individual’s health is typically referred to as personalized or companion diagnostics. Furthermore, once the molecular mechanism of the mutations is revealed, the patient should be given drugs which are the most appropriate for the individual genome, referred to as pharmacogenomics. Altogether, the shift in focus in medicine towards more genomic-oriented practices is the foundation of personalized medicine. The progress made in these rapidly developing fields is outlined.

Exploring the function of genetic variants in the non-coding genomic regions: approaches for identifying human regulatory variants affecting gene expression

Understanding the genetic basis of human traits/diseases and the underlying mechanisms of how these traits/diseases are affected by genetic variations is critical for public health. Current genome-wide functional genomics data uncovered a large number of functional elements in the noncoding regions of human genome, providing new opportunities to study regulatory variants (RVs). RVs play important roles in transcription factor bindings, chromatin states and epigenetic modifications. Here, we systematically review an array of methods currently used to map RVs as well as the computational approaches in annotating and interpreting their regulatory effects, with emphasis on regulatory single-nucleotide polymorphism. We also briefly introduce experimental methods to validate these functional RVs.

Production and analytic bioinformatics for next-generation DNA sequencing

The bioinformatics requirements within the clinical environment are very specific, and analytic techniques need to be fit for purpose, robust, and predictable. At the same time, the bewildering amount of information produced during these analyses needs to be carefully managed, used and interpreted correctly. The challenge for clinical laboratories now is to implement production analytical processes that are capable of handling different experimental approaches on current equipment, as well as to incorporate ways for these systems to evolve to take account of developments likely to make impacts in the near future. This is complicated by the many options available at each of the critical processing steps and a clear method needs to be developed to assemble appropriate pipelines. Here, I discuss the issues relevant to the development of an informatics pipeline that meets these criteria that should allow individual laboratories to assess their proposed strategies.

A primer for disease gene prioritization using next-generation sequencing data

High-throughput next-generation sequencing (NGS) technology produces a tremendous amount of raw sequence data. The challenges for researchers are to process the raw data, to map the sequences to genome, to discover variants that are different from the reference genome, and to prioritize/rank the variants for the question of interest. The recent development of many computational algorithms and programs has vastly improved the ability to translate sequence data into valuable information for disease gene identification. However, the NGS data analysis is complex and could be overwhelming for researchers who are not familiar with the process. Here, we outline the analysis pipeline and describe some of the most commonly used principles and tools for analyzing NGS data for disease gene identification.

Profiling, Bioinformatic, and Functional Data on the Developing Olfactory/GnRH System Reveal Cellular and Molecular Pathways Essential for This Process and Potentially Relevant for the Kallmann Syndrome

During embryonic development, immature neurons in the olfactory epithelium (OE) extend axons through the nasal mesenchyme, to contact projection neurons in the olfactory bulb. Axon navigation is accompanied by migration of the GnRH+ neurons, which enter the anterior forebrain and home in the septo-hypothalamic area. This process can be interrupted at various points and lead to the onset of the Kallmann syndrome (KS), a disorder characterized by anosmia and central hypogonadotropic hypogonadism. Several genes has been identified in human and mice that cause KS or a KS-like phenotype. In mice a set of transcription factors appears to be required for olfactory connectivity and GnRH neuron migration; thus we explored the transcriptional network underlying this developmental process by profiling the OE and the adjacent mesenchyme at three embryonic ages. We also profiled the OE from embryos null for Dlx5, a homeogene that causes a KS-like phenotype when deleted. We identified 20 interesting genes belonging to the following categories: (1) transmembrane adhesion/receptor, (2) axon-glia interaction, (3) scaffold/adapter for signaling, (4) synaptic proteins. We tested some of them in zebrafish embryos: the depletion of five (of six) Dlx5 targets affected axonal extension and targeting, while three (of three) affected GnRH neuron position and neurite organization. Thus, we confirmed the importance of cell-cell and cell-matrix interactions and identified new molecules needed for olfactory connection and GnRH neuron migration. Using available and newly generated data, we predicted/prioritized putative KS-disease genes, by building conserved co-expression networks with all known disease genes in human and mouse. The results show the overall validity of approaches based on high-throughput data and predictive bioinformatics to identify genes potentially relevant for the molecular pathogenesis of KS. A number of candidate will be discussed, that should be tested in future mutation screens.