[Show abstract][Hide abstract]ABSTRACT: The Bioinformatics book covers new topics in the rapidly expanding field of bioinformatics, from next-generation sequencing to drug discovery and metagenomics.
The first two chapters overviews genetic measurement methods. The next four chapters discuss topics related to the effect of genetic variants from protein modeling to gene regulatory networks. Standard statistical analysis in association studies are discussed in the next two chapters. The systems biology approach is illustrated by discussing a systems-based biomarker analysis method, the graph-based network science, the dynamical systems based approaches and a Bayesian causal inference method in subsequent chapters. The next chapter discusses text-mining methods in biomedicine, especially their application in interpretation and translation. The decision theoretic approach to study design, especially multi-stage, sequential study design is discussed in the next chapter, introducing the concepts of value of information and the expected value of an experiment. Next, the heterogeneity of biomedical big data sources is overviewed, together with data and knowledge fusion methods, and with the discussion of semantic publishing, which can lead to a new unification of biomedicine. Subsequently, bioinformatic workflow methods are summarized. At last, drug discovery methods are overviewed with an outlook for personalized medicine and the final chapter presents the main steps and workflows in metagenomics.
[Show abstract][Hide abstract]ABSTRACT: Haplotype blocks of the investigated SNPs in the present study. SNPs are numbered sequentially and their relative location is indicated along the top. Markers 1–54 and markers 55–102 correspond to the studied SNPs on Chromosome 14 and 11, respectively. Triangles surrounding markers represent haplotype blocks.
[Show abstract][Hide abstract]ABSTRACT: Comparison of standard concept of (pairwise) association and strong relevance. The concept of strong relevance and association (with respect to a target) has only one common element, direct relevance, i.e. non-mediated relationship between the target and a variable. Association also includes confounded and transitive relevance, where there is a mediator between the given variable and the target. In cause-effect relationship terms, the confounded case corresponds to a common cause; the transitive case corresponds to a cause- effect path. Strong relevance, on the other hand, includes interactionist relevance, i.e. a common effect type relationship.
[Show abstract][Hide abstract]ABSTRACT: Genetic studies indicate high number of potential factors related to asthma. Based on earlier linkage analyses we selected the 11q13 and 14q22 asthma susceptibility regions, for which we designed a partial genome screening study using 145 SNPs in 1201 individuals (436 asthmatic children and 765 controls). The results were evaluated with traditional frequentist methods and we applied a new statistical method, called Bayesian network based Bayesian multilevel analysis of relevance (BN-BMLA). This method uses Bayesian network representation to provide detailed characterization of the relevance of factors, such as joint significance, the type of dependency, and multi-target aspects. We estimated posteriors for these relations within the Bayesian statistical framework, in order to estimate the posteriors whether a variable is directly relevant or its association is only mediated.
With frequentist methods one SNP (rs3751464 in the FRMD6 gene) provided evidence for an association with asthma (OR = 1.43(1.2–1.8); p = 3×10−4). The possible role of the FRMD6 gene in asthma was also confirmed in an animal model and human asthmatics.
In the BN-BMLA analysis altogether 5 SNPs in 4 genes were found relevant in connection with asthma phenotype: PRPF19 on chromosome 11, and FRMD6, PTGER2 and PTGDR on chromosome 14. In a subsequent step a partial dataset containing rhinitis and further clinical parameters was used, which allowed the analysis of relevance of SNPs for asthma and multiple targets. These analyses suggested that SNPs in the AHNAK and MS4A2 genes were indirectly associated with asthma. This paper indicates that BN-BMLA explores the relevant factors more comprehensively than traditional statistical methods and extends the scope of strong relevance based methods to include partial relevance, global characterization of relevance and multi-target relevance.
[Show abstract][Hide abstract]ABSTRACT: The most probable univariate (MBM), bivariate (2-MBS), trivariate (3-MBS), 4-MBS, 5-MBS subsets of variables. All posteriors of partial k-relevance are ordered, x axis denotes the rank (e.g.: the number 2 means the second highest posterior), and y axis denotes the posterior probabilities. As k increases (i.e. the set size of jointly considered SNPs) the maximum posteriors decrease, and the slope of the posteriors are much less “peaked”, which means that the lower ranked posteriors are significantly higher than those with higher ranks. The univariate case can be considered as a relatively peaked curve, and as k increases, the curve “flattens”. The curve of whole sets (MBS) is the “flattest”, showing only a small difference between the lower ranked and the higher ranked posteriors. In this case, estimating the top 20 MBS is less informative than estimating the top 10 partial k-relevance of k = 2. This indicates the viability of the concept of partial k-relevance.
[Show abstract][Hide abstract]ABSTRACT: The peakness of the posteriors of the most probable 100 MBS sets. The x axis denotes the rank of a Markov blanket set (MBS) of SNPs, the y axis denotes the joint probability of an MBS, e.g. an MBS with rank = 5 is the fifth most probable set. The “RA” and “CLI” prefixes denote the corresponding dataset, and the “asthma” and “multitarget” suffixes indicate the target variables. RA- multitarget: Rhinitis, Asthma; CLI- multitarget: Asthma, Rhinitis, Eosinophil and IgE level. Note, that the peakness of the posteriors decreases within the same dataset in the multitarget case; and between different data sets, the smaller sample size (CLI:200, RA:1100) results in weaker posteriors. In terms of data sufficiency, the flatness of the CLI MBS posterior curve (both in the single target and the multitarget case) indicates that the CLI data is not sufficient for a complete multivariate analysis. The RA dataset, on the other hand, is more appropriate having a relatively peaked posterior, although the maximum posterior is not particularly high.
[Show abstract][Hide abstract]ABSTRACT: Psychoneuroimmunologic studies on positive emotions are few, and their clinical relevance is limited.
This "SHoRT" (Smiling Hospital Research Team) study evaluates the effects that Smiling Hospital artists have on hospitalized children.
Blood samples were taken in a non-painful way through branules in an accredited Infectology Ward, 30 minutes before and 1 hour after a visit of tale tellers, puppeteers and handicraft artists. 24 children were visited and 9 were included in the control group. Blood lymphocyte counts and Th1/Th2 cytokine levels were determined. Artists evaluated their effect on a subjective scale.
In the visited group, the increase of lymphocytes was 8.43% higher, the decrease was 12.45% lower, and the proportion of children showing increased lymphocyte counts was more increased. Changes were more marked after more successful visits. Authors found non-significant, still considerable changes in interferon-γ level (p < 0.055) and in Th1/Th2 cytokine ratios.
This pediatric study suggests that immunological changes may develop when more attention is given to hospitalized children.
[Show abstract][Hide abstract]ABSTRACT: The accumulation of electronically accessible data and knowledge are posing theoretical and practical challenges for study
design and statistical data analysis. It consists of the use of the results of earlier high-throughput measurements of genetic
variations, microRNA, and gene expression levels, and the use of the biological knowledge bases. We investigate fusion in
the phases of study design, data analysis, and interpretation; specifically, we present methodologies and bioinformatic tools
in the Bayesian framework to deepen, lengthen, and broaden this fusion. First, we overview a Bayesian decision support for
design of partial genetic association studies (GASs) incorporating domain literature, knowledge bases, and results of analysis
of earlier studies. Second, we present a Bayesian multilevel analysis (BMLA) for GAS, which performs an integrated analysis
at the univariate and multivariate levels, and at the level of interactions. Third, we present a Bayesian logic to support
interpretation, which integrates the results of data analysis and factual domain knowledge. Finally, we discuss the advantages
of the Bayesian framework to cope with small sample size, fusion of data and knowledge, challenges of multiple testing, meta-analysis,
and positive results bias (i.e., the communication of scientific uncertainty). The genomics of asthma will serve as an application