Marcin Kierczak

Genome-Wide Analysis in German Shepherd Dogs Reveals Association of a Locus on CFA 27 with Atopic Dermatitis

Humans and dogs are both affected by the allergic skin disease atopic dermatitis (AD), caused by an interaction between genetic and environmental factors. The German shepherd dog (GSD) is a high-risk breed for canine AD (CAD). In this study, we used a Swedish cohort of GSDs as a model for human AD. Serum IgA levels are known to be lower in GSDs compared to other breeds. We detected significantly lower IgA levels in the CAD cases compared to controls (p = 1.1×10−5) in our study population. We also detected a separation within the GSD cohort, where dogs could be grouped into two different subpopulations. Disease prevalence differed significantly between the subpopulations contributing to population stratification (λ = 1.3), which was successfully corrected for using a mixed model approach. A genome-wide association analysis of CAD was performed (n cases = 91, n controls = 88). IgA levels were included in the model, due to the high correlation between CAD and low IgA levels. In addition, we detected a correlation between IgA levels and the age at the time of sampling (corr = 0.42, p = 3.0×10−9), thus age was included in the model. A genome-wide significant association was detected on chromosome 27 (praw = 3.1×10−7, pgenome = 0.03). The total associated region was defined as a ∼1.5-Mb-long haplotype including eight genes. Through targeted re-sequencing and additional genotyping of a subset of identified SNPs, we defined 11 smaller haplotype blocks within the associated region. Two blocks showed the strongest association to CAD. The ∼209-kb region, defined by the two blocks, harbors only the PKP2 gene, encoding Plakophilin 2 expressed in the desmosomes and important for skin structure. Our results may yield further insight into the genetics behind both canine and human AD.

Fine‐grained adaptive divergence in an amphibian: genetic basis of phenotypic divergence and the role of nonrandom gene flow in restricting effective migration among wetlands

Adaptive ecological differentiation among sympatric populations is promoted by environmental heterogeneity, strong local selection and restricted gene flow. High gene flow, on the other hand, is expected to homogenize genetic variation among populations and therefore prevent local adaptation. Understanding how local adaptation can persist at the spatial scale at which gene flow occurs has remained an elusive goal, especially for wild vertebrate populations. Here, we explore the roles of natural selection and nonrandom gene flow (isolation by breeding time and habitat choice) in restricting effective migration among local populations and promoting generalized genetic barriers to neutral gene flow. We examined these processes in a network of 17 breeding ponds of the moor frog Rana arvalis, by combining environmental field data, a common garden experiment and data on variation in neutral microsatellite loci and in a thyroid hormone receptor (TRβ) gene putatively under selection. We illustrate the connection between genotype, phenotype and habitat variation and demonstrate that the strong differences in larval life history traits observed in the common garden experiment can result from adaptation to local pond characteristics. Remarkably, we found that haplotype variation in the TRβ gene contributes to variation in larval development time and growth rate, indicating that polymorphism in the TRβ gene is linked with the phenotypic variation among the environments. Genetic distance in neutral markers was correlated with differences in breeding time and environmental differences among the ponds, but not with geographical distance. These results demonstrate that while our study area did not exceed the scale of gene flow, ecological barriers constrained gene flow among contrasting habitats. Our results highlight the roles of strong selection and nonrandom gene flow created by phenological variation and, possibly, habitat preferences, which together maintain genetic and phenotypic divergence at a fine‐grained spatial scale.

Breed Differences in Natriuretic Peptides in Healthy Dogs

Background Measurement of plasma concentration of natriuretic peptides (NPs) is suggested to be of value in diagnosis of cardiac disease in dogs, but many factors other than cardiac status may influence their concentrations. Dog breed potentially is 1 such factor. Objective To investigate breed variation in plasma concentrations of pro‐atrial natriuretic peptide 31‐67 (proANP 31‐67) and N‐terminal B‐type natriuretic peptide (NT‐proBNP) in healthy dogs. Animals 535 healthy, privately owned dogs of 9 breeds were examined at 5 centers as part of the European Union (EU) LUPA project. Methods Absence of cardiovascular disease or other clinically relevant organ‐related or systemic disease was ensured by thorough clinical investigation. Plasma concentrations of proANP 31‐67 and NT‐proBNP were measured by commercially available ELISA assays. Results Overall significant breed differences were found in proANP 31‐67 (P < .0001) and NT‐proBNP (P < .0001) concentrations. Pair‐wise comparisons between breeds differed in approximately 50% of comparisons for proANP 31‐67 as well as NT‐proBNP concentrations, both when including all centers and within each center. Interquartile range was large for many breeds, especially for NT‐proBNP. Among included breeds, Labrador Retrievers and Newfoundlands had highest median NT‐proBNP concentrations with concentrations 3 times as high as those of Dachshunds. German Shepherds and Cavalier King Charles Spaniels had the highest median proANP 31‐67 concentrations, twice the median concentration in Doberman Pinschers. Conclusions and Clinical Importance Considerable interbreed variation in plasma NP concentrations was found in healthy dogs. Intrabreed variation was large in several breeds, especially for NT‐proBNP. Additional studies are needed to establish breed‐specific reference ranges.

A Rough Set-Based Model of HIV-1 Reverse Transcriptase Resistome

Reverse transcriptase (RT) is a viral enzyme crucial for HIV-1 replication. Currently, 12 drugs are targeted against the RT. The low fidelity of the RT-mediated transcription leads to the quick accumulation of drug-resistance mutations. The sequence-resistance relationship remains only partially understood. Using publicly available data collected from over 15 years of HIV proteome research, we have created a general and predictive rule-based model of HIV-1 resistance to eight RT inhibitors. Our rough set-based model considers changes in the physicochemical properties of a mutated sequence as compared to the wild-type strain. Thanks to the application of the Monte Carlo feature selection method, the model takes into account only the properties that significantly contribute to the resistance phenomenon. The obtained results show that drug-resistance is determined in more complex way than believed. We confirmed the importance of many resistance-associated sites, found some sites to be less relevant than formerly postulated and—more importantly—identified several previously neglected sites as potentially relevant. By mapping some of the newly discovered sites on the 3D structure of the RT, we were able to suggest possible molecular-mechanisms of drug-resistance. Importantly, our model has the ability to generalize predictions to the previously unseen cases. The study is an example of how computational biology methods can increase our understanding of the HIV-1 resistome.

Ensemble learning prediction of protein–protein interactions using proteins functional annotations

Protein-protein interactions are important for the majority of biological processes. A significant number of computational methods have been developed to predict protein-protein interactions using protein sequence, structural and genomic data. Vast experimental data is publicly available on the Internet, but it is scattered across numerous databases. This fact motivated us to create and evaluate new high-throughput datasets of interacting proteins. We extracted interaction data from DIP, MINT, BioGRID and IntAct databases. Then we constructed descriptive features for machine learning purposes based on data from Gene Ontology and DOMINE. Thereafter, four well-established machine learning methods: Support Vector Machine, Random Forest, Decision Tree and Naïve Bayes, were used on these datasets to build an Ensemble Learning method based on majority voting. In cross-validation experiment, sensitivity exceeded 80% and classification/prediction accuracy reached 90% for the Ensemble Learning method. We extended the experiment to a bigger and more realistic dataset maintaining sensitivity over 70%. These results confirmed that our datasets are suitable for performing PPI prediction and Ensemble Learning method is well suited for this task. Both the processed PPI datasets and the software are available at .

Identification of candidate genes and mutations in QTL regions for chicken growth using bioinformatic analysis of NGS and SNP-chip data

Mapping of chromosomal regions harboring genetic polymorphisms that regulate complex traits is usually followed by a search for the causative mutations underlying the observed effects. This is often a challenging task even after fine mapping, as millions of base pairs including many genes will typically need to be investigated. Thus to trace the causative mutation(s) there is a great need for efficient bioinformatic strategies. Here, we searched for genes and mutations regulating growth in the Virginia chicken lines – an experimental population comprising two lines that have been divergently selected for body weight at 56 days for more than 50 generations. Several quantitative trait loci (QTL) have been mapped in an F2 intercross between the lines, and the regions have subsequently been replicated and fine mapped using an Advanced Intercross Line. We have further analyzed the QTL regions where the largest genetic divergence between the High-Weight selected (HWS) and Low-Weight selected (LWS) lines was observed. Such regions, covering about 37% of the actual QTL regions, were identified by comparing the allele frequencies of the HWS and LWS lines using both individual 60K SNP chip genotyping of birds and analysis of read proportions from genome resequencing of DNA pools. Based on a combination of criteria including significance of the QTL, allele frequency difference of identified mutations between the selected lines, gene information on relevance for growth, and the predicted functional effects of identified mutations we propose here a subset of candidate mutations of highest priority for further evaluation in functional studies. The candidate mutations were identified within the GCG, IGFBP2, GRB14, CRIM1, FGF16, VEGFR-2, ALG11, EDN1, SNX6, and BIRC7 genes. We believe that the proposed method of combining different types of genomic information increases the probability that the genes underlying the observed QTL effects are represented among the candidate mutations identified.

Monte Carlo feature selection and interdependency discovery in supervised classification

Applications of machine learning techniques in Life Sciences are the main applications forcing a paradigm shift in the way these techniques are used. Rather than obtaining the best possible supervised classifier, the Life Scientist needs to know which features contribute best to classifying observations into distinct classes and what are the interdependencies between the features. To this end we significantly extend our earlier work [Draminski et al. (2008)] that introduced an effective and reliable method for ranking features according to their importance for classification. We begin with adding a method for finding a cut-off between informative and non-informative features and then continue with a development of a methodology and an implementation of a procedure for determining interdependencies between informative features. The reliability of our approach rests on multiple construction of tree classifiers. Essentially, each classifier is trained on a randomly chosen subset of the original data using only a fraction of all of the observed features. This approach is conceptually simple yet computer-intensive. The methodology is validated on a large and difficult task of modelling HIV-1 reverse transcriptase resistance to drugs which is a good example of the aforementioned paradigm shift. In this task, of the main interest is the identification of mutation points (i.e. features) and their combinations that model drug resistance.

Two loci on chromosome 5 are associated with serum IgE levels in Labrador retrievers.

Crosslinking of immunoglobulin E antibodies (IgE) bound at the surface of mast cells and subsequent mediator release is considered the most important trigger for allergic reactions. Therefore, the genetic control of IgE levels is studied in the context of allergic diseases, such as asthma, atopic rhinitis, or atopic dermatitis (AD). We performed genome-wide association studies in 161 Labrador Retrievers with regard to total and allergen-specific immunoglobulin E (IgE) levels. We identified a genome-wide significant association on CFA 5 with the antigen-specific IgE responsiveness to Acarus siro. We detected a second genome-wide significant association with respect to the antigen-specific IgE responsiveness to Tyrophagus putrescentiae at a different locus on chromosome 5. A. siro and T. putrescentiae both belong to the family Acaridae and represent so-called storage or forage mites. These forage mites are discussed as major allergen sources in canine AD. No obvious candidate gene for the regulation of IgE levels is located under the two association signals. Therefore our studies offer a chance of identifying a novel mechanism controlling the hosts IgE response.

PASE: a novel method for functional prediction of amino acid substitutions based on physicochemical properties.

Background: Non-synonymous single-nucleotide polymorphisms (nsSNPs) within the coding regions of genes causing amino acid substitutions (AASs) may have a large impact on protein function. The possibilities to identify nsSNPs across genomes have increased notably with the advent of next-generation sequencing technologies. Thus, there is a strong need for efficient bioinformatics tools to predict the functional effect of AASs. Such tools can be used to identify the most promising candidate mutations for further experimental validation. Results: Here we present prediction of AAS effects (PASE), a novel method that predicts the effect of an AASs based on physicochemical property changes. Evaluation of PASE, using a few AASs of known phenotypic effects and 3338 human AASs, for which functional effects have previously been scored with the widely used SIFT and PolyPhen tools, show that PASE is a useful method for functional prediction of AASs. We also show that the predictions can be further improved by combining PASE with information about evolutionary conservation. Conclusion: PASE is a novel algorithm for predicting functional effects of AASs, which can be used for pinpointing the most interesting candidate mutations. PASE predictions are based on changes in seven physicochemical properties and can improve predictions from many other available tools, which are based on evolutionary conservation. Using available experimental data and predictions from the already existing tools, we demonstrate that PASE is a useful method for predicting functional effects of AASs, even when a limited number of query sequence homologs/orthologs are available.

Computational Analysis of Molecular Interaction Networks Underlying Change of HIV-1 Resistance to Selected Reverse Transcriptase Inhibitors

Motivation Despite more than two decades of research, HIV resistance to drugs remains a serious obstacle in developing efficient AIDS treatments. Several computational methods have been developed to predict resistance level from the sequence of viral proteins such as reverse transcriptase (RT) or protease. These methods, while powerful and accurate, give very little insight into the molecular interactions that underly acquisition of drug resistance/hypersusceptibility. Here, we attempt at filling this gap by using our Monte Carlo feature selection and interdependency discovery method (MCFS-ID) to elucidate molecular interaction networks that characterize viral strains with altered drug resistance levels. Results We analyzed a number of HIV-1 RT sequences annotated with drug resistance level using the MCFS-ID method. This let us expound interdependency networks that characterize change of drug resistance to six selected RT inhibitors: Abacavir, Lamivudine, Stavudine, Zidovudine, Tenofovir and Nevirapine. The networks consider interdependencies at the level of physicochemical properties of mutating amino acids, eg,: polarity. We mapped each network on the 3D structure of RT in attempt to understand the molecular meaning of interacting pairs. The discovered interactions describe several known drug resistance mechanisms and, importantly, some previously unidentified ones. Our approach can be easily applied to a whole range of problems from the domain of protein engineering. Availability A portable Java implementation of our MCFS-ID method is freely available for academic users and can be obtained at: http://www.ipipan.eu/staff/m.draminski/software.htm.