Clinical and pharmacogenomic data mining: 2. A simple method for the combination of information from associations and multivariances to facilitate analysis, decision, and design in clinical research and practice

The physician and researcher must ultimately be able to combine qualitative and quantitative features from a variety of combinations of observations on data of many component items (i.e., many dimensions), and hence reach simple conclusions about interpretation, rational courses of action, and design. In the first paper of this series, it was noted that such needs are challenging the classical means of using statistics. Hence, the paper proposed the use of a Generalized Theory of Expected Information or "Zeta Theory". The conjoint event [a,b,c,..] is seen as a rule of association for a,b,c,.. associated with a rule strength I(a;b;c;...) = xi(s,o[a,b,c,..]) - xi (s,e[a,b,c,...]), where xi is the incomplete Zeta Function. Here, o[a,b,c,...] is the observed, and e[a,b,c,..] the expected, frequency of occurrence of conjoint event [a,b,c,...]. The present paper explores how output from this approach might be assembled in a form better suited for decision support. Related to this is the difficulty that the treatment of covariance and multivariance was previously rendered as a "fuzzy association" so that the output would fall into a similar form as the true associations, but this was a somewhat ad hoc approach in which only the final I( ) had any meaning. Users at clinical research sites had subsequently requested an alternative approach in which "effective frequencies" o[ ] and e[ ] calculated from the above variances and used to evaluate I( ) give some intuitive feeling analogous to the association treatment, and this is explored here. Though the present paper is theoretical, real examples are used to illustrate application. One clinical-genomic example illustrates experimental design by identifying data which is, or is not, statistically germane to the study. We also report on some impressions based on applying these techniques in studies of real, extensive patient record data which are now emerging, as well as on molecular design data originally studied in part to test the ability to deduce the effects of simple natural patient sequence variations ("SNPs") on patient protein activity. On the basis of these study experiences, methods of rationalizing and condensing the rules implied by associations and variances between data, as well as discussion of the difficulty of what is meant by "condensed", are presented in the Appendix.     

Requirement of chondroitin sulfate/dermatan sulfate recognition in midkine-dependent migration of macrophages

Midkine (MK) is a heparin-binding growth factor that promotes cell migration, cell growth and cell survival. The promotion of migration of inflammatory cells, especially macrophages, by MK is involved in formation of a vascular abnormality, i.e. neointima formation. MK-induced migration of peritoneal exudate macrophages was inhibited by heparin, chondroitin sulfate E and dermatan sulfate, but not by chondroitin sulfate D or chondroitin 6-sulfate. Digestion of macrophages with chondroitinase ABC as well as chondroitinase B decreased the migratory activity. However, heparitinase digestion showed only slight effects. These results indicated that a chondroitin sulfate, i.e. an E-type oversulfated structure with dermatan sulfate domain, is involved in MK-induced migration of macrophages. Although a chondroitin sulfate proteoglycan, receptor-type protein tyrosine phosphatase (PTP ), participates in MK-induced migration of neurons and osteoblasts, PTP was not detected in macrophages. The MK-induced migration was inhibited by PP1, wortomanin, PD 98059 and vanadate, indicating that the downstream signaling system, which includes Src, PI3 kinase and ERK as important components, is shared with other MK signaling systems in which PTP is involved.     

Determination of zeta potential in Planctomycetes and its application in heavy metals toxicity assessment

Zeta potential of Planctomycetes was evaluated under different environmental conditions and correlated to cell viability. Phylogenetically distinct strains of the Planctomycetes presented different negative zeta potential values. More negative values were associated with Rhodopirellula spp. and related to the great amount of fimbriae in these species. Milli-Q water was chosen as the best dispersion media to perform the measurements. Zeta potential increased with ionic strength and varied with pH. In the physiological range of pH 5.0-9, zeta potential remained low and Rhodopirellula sp. strain LF2 cells were viable. Out of this range, zeta potential increased significantly and viability decreased. The effect on zeta potential of arsenic, cadmium, chromium, copper, lead, nickel, and zinc was assessed in Rhodopirellula sp. strain LF2. Zeta potential increased with increasing toxicity of the heavy metals in a dose-response way. This result was confirmed by the results observed for Rhodopirellula baltica strain SH1 under copper toxicity. Lead was the most toxic metal and zinc was the least toxic as observed by zeta potential and viability. The results support a correlation between zeta potential and cell viability which seem to indicate the possibility to use it as a viability predictor for the effects of heavy metals toxicity.     

The new physician as unwitting quantum mechanic: is adapting Dirac's inference system best practice for personalized medicine, genomics, and proteomics?

What is the Best Practice for automated inference in Medical Decision Support for personalized medicine? A known system already exists as Dirac's inference system from quantum mechanics (QM) using bra-kets and bras where A and B are states, events, or measurements representing, say, clinical and biomedical rules. Dirac's system should theoretically be the universal best practice for all inference, though QM is notorious as sometimes leading to bizarre conclusions that appear not to be applicable to the macroscopic world of everyday world human experience and medical practice. It is here argued that this apparent difficulty vanishes if QM is assigned one new multiplication function @, which conserves conditionality appropriately, making QM applicable to classical inference including a quantitative form of the predicate calculus. An alternative interpretation with the same consequences is if every i = radical-1 in Dirac's QM is replaced by h, an entity distinct from 1 and i and arguably a hidden root of 1 such that h2 = 1. With that exception, this paper is thus primarily a review of the application of Dirac's system, by application of linear algebra in the complex domain to help manipulate information about associations and ontology in complicated data. Any combined bra-ket can be shown to be composed only of the sum of QM-like bra and ket weights c(), times an exponential function of Fano's mutual information measure I(A; B) about the association between A and B, that is, an association rule from data mining. With the weights and Fano measure re-expressed as expectations on finite data using Riemann's Incomplete (i.e., Generalized) Zeta Functions, actual counts of observations for real world sparse data can be readily utilized. Finally, the paper compares identical character, distinguishability of states events or measurements, correlation, mutual information, and orthogonal character, important issues in data mining and biomedical analytics, as in QM.     

A Bayesian Alternative to Mutual Information for the Hierarchical Clustering of Dependent Random Variables

The use of mutual information as a similarity measure in agglomerative hierarchical clustering (AHC) raises an important issue: some correction needs to be applied for the dimensionality of variables. In this work, we formulate the decision of merging dependent multivariate normal variables in an AHC procedure as a Bayesian model comparison. We found that the Bayesian formulation naturally shrinks the empirical covariance matrix towards a matrix set a priori (e.g., the identity), provides an automated stopping rule, and corrects for dimensionality using a term that scales up the measure as a function of the dimensionality of the variables. Also, the resulting log Bayes factor is asymptotically proportional to the plug-in estimate of mutual information, with an additive correction for dimensionality in agreement with the Bayesian information criterion. We investigated the behavior of these Bayesian alternatives (in exact and asymptotic forms) to mutual information on simulated and real data. An encouraging result was first derived on simulations: the hierarchical clustering based on the log Bayes factor outperformed off-the-shelf clustering techniques as well as raw and normalized mutual information in terms of classification accuracy. On a toy example, we found that the Bayesian approaches led to results that were similar to those of mutual information clustering techniques, with the advantage of an automated thresholding. On real functional magnetic resonance imaging (fMRI) datasets measuring brain activity, it identified clusters consistent with the established outcome of standard procedures. On this application, normalized mutual information had a highly atypical behavior, in the sense that it systematically favored very large clusters. These initial experiments suggest that the proposed Bayesian alternatives to mutual information are a useful new tool for hierarchical clustering.     

Characterization of Bacillus subtilis clones surviving overproduction of Zeta, a pSM19035 plasmid-encoded toxin

The postsegregational killing system of pSM19035 plasmid consists of the proteins Zeta and Epsilon, a toxin and an antidote, respectively. Zeta mutants were isolated with the use of Bacillus subtilis strain with the zeta gene under control of an inducible promoter integrated into the chromosome. Results of mutant analysis point to the amino terminal part of the Zeta protein as being responsible for the toxicity.     

Structural model of the transmembrane Fo rotary sector of H+-transporting ATP synthase derived by solution NMR and intersubunit cross-linking in situ

H(+)-transporting, F(1)F(o)-type ATP synthases utilize a transmembrane H(+) potential to drive ATP formation by a rotary catalytic mechanism. ATP is formed in alternating beta subunits of the extramembranous F(1) sector of the enzyme, synthesis being driven by rotation of the gamma subunit in the center of the F(1) molecule between the alternating catalytic sites. The H(+) electrochemical potential is thought to drive gamma subunit rotation by first coupling H(+) transport to rotation of an oligomeric rotor of c subunits within the transmembrane F(o) sector. The gamma subunit is forced to turn with the c-oligomeric rotor due to connections between subunit c and the gamma and epsilon subunits of F(1). In this essay we will review recent studies on the Escherichia coli F(o) sector. The monomeric structure of subunit c, determined by NMR, shows that subunit c folds in a helical hairpin with the proton carrying Asp(61) centered in the second transmembrane helix (TMH). A model for the structural organization of the c(10) oligomer in F(o) was deduced from extensive cross-linking studies and by molecular modeling. The model indicates that the H(+)-carrying carboxyl of subunit c is occluded between neighboring subunits of the c(10) oligomer and that two c subunits pack in a "front-to-back" manner to form the H(+) (cation) binding site. In order for protons to gain access to Asp(61) during the protonation/deprotonation cycle, we propose that the outer, Asp(61)-bearing TMH-2s of the c-ring and TMHs from subunits composing the inlet and outlet channels must turn relative to each other, and that the swiveling motion associated with Asp(61) protonation/deprotonation drives the rotation of the c-ring. The NMR structures of wild-type subunit c differs according to the protonation state of Asp(61). The idea that the conformational state of subunit c changes during the catalytic cycle is supported by the cross-linking evidence in situ, and two recent NMR structures of functional mutant proteins in which critical residues have been switched between TMH-1 and TMH-2. The structural information is considered in the context of the possible mechanism of rotary movement of the c(10) oligomer during coupled synthesis of ATP.     

Insulin induces phosphorylation of the AMPA receptor subunit GluR1, reversed by ZIP,and over‐expression of Protein Kinase M zeta,reversed by amyloid beta

Insulin receptor (IR) in the brain plays a role in synaptic plasticity and cognitive functions. Phosphorylation of α‐amino‐3‐hydroxy‐5‐methylisoxazole‐4‐propionic acid (AMPA) receptors GluR1 subunit at Serine 831 is regulated by calcium–calmodulin‐dependent protein kinase II and protein kinase C that underlie long‐term potentiation and learning/memory. Recent studies have shown that the novel Protein Kinase M zeta (PKMζ) underlies synaptic plasticity and may regulate AMPAr. In this study, we show that insulin induces phosphorylation of Serine 831 GluR1 subunit of AMPAr and induces over‐expression of PKMζ; pre‐treatment with either the IR inhibitor 3‐Bromo‐5‐t‐butyl‐4‐hydroxy‐benzylidenemalonitrile (AG1024) or PKMζ inhibitor protein kinase C zeta pseudo‐substrate inhibitor returned the phosphorylation value of GluR1 to control level. Amyloid beta (Aβ) peptide in the form of oligomers interferes with IR signaling. Pre‐treating neuronal cultures with Aβ following incubation with insulin, we found a reduction of insulin‐dependent PKMζ over‐expression and MAPK/Erk (1/2) phosphorylation, i.e., signaling pathways involved in synaptic plasticity and learning/memory. These results indicate a new intracellular insulin signaling pathway, and, additionally, that insulin resistance in Alzheimer's disease is a response to the production and accumulation of Aβ.

     

Active site control of myosin cross-bridge zeta potential

The electrical properties of contractile proteins contribute to muscle structure and perhaps function but have not been characterized adequately. Electrophoretic mobility, mu(e), is sensitive to the net electric charge and hydrodynamic size of a molecule in solution. Zeta potential, zeta, particle charge, Q(e), and particle charge-to-mass ratio are proportional to mu(e). We measured mu(e) for nucleotide complexes of skeletal muscle heavy meromyosin (HMM) and subfragment 1 (S1). The results indicate that mu(e) for HMM changes depending on the ligand bound in the active site. The changes in electric charge appear to occur mainly on the S1 moieties. For HMM(MgATPgammaS)(2) and HMM(MgADP.P(i))(2) the values of mu(e) are -0.077 and -0.17 (microm/s)/(V/cm), respectively. For these complexes, mu(e) is independent of [ATP], [ADP], and [P(i)]. When P(i) dissociates from HMM(MgADP.P(i))(2) to form HMM(MgADP)(2), mu(e) decreases to -0.61 (microm/s)/(V/cm). This large decrease in mu(e) is independent of free [ADP] or [ATP]. Increasing [P(i)], on the other hand, increases mu(e) for HMM(MgADP)(2) to values near those observed for the steady-state intermediate. For HMM, mu(e) = -0.34 and is independent of P(i). MgADP binding to HMM decreases mu(e) to -0.57 (microm/s)/(V/cm), and the dissociation constant is 9 microM. Taken together, these data indicate that mu(e) and, thus, zeta are controlled by ligand binding to the active site. The magnitudes of the particle charge-to-mass ratios for the HMM complexes are all in a range that falls within published values determined for a variety of other proteins. Possible roles that the observed nucleotide-dependent changes in cross-bridge electric charge might have in the contractile cycle in muscle are considered.     

Incorporating prior information with fused sparse group lasso: Application to prediction of clinical measures from neuroimages

Predicting clinical variables from whole‐brain neuroimages is a high‐dimensional problem that can potentially benefit from feature selection or extraction. Penalized regression is a popular embedded feature selection method for high‐dimensional data. For neuroimaging applications, spatial regularization using the or norm of the image gradient has shown good performance, yielding smooth solutions in spatially contiguous brain regions. Enormous resources have been devoted to establishing structural and functional brain connectivity networks that can be used to define spatially distributed yet related groups of voxels. We propose using the fused sparse group lasso (FSGL) penalty to encourage structured, sparse, and interpretable solutions by incorporating prior information about spatial and group structure among voxels. We present optimization steps for FSGL penalized regression using the alternating direction method of multipliers algorithm. With simulation studies and in application to real functional magnetic resonance imaging data from the Autism Brain Imaging Data Exchange, we demonstrate conditions under which fusion and group penalty terms together outperform either of them alone.     

Application of novel and existing methods to identify genes with evidence of epigenetic association: results from GAW20

Background

The rise in popularity and accessibility of DNA methylation data to evaluate epigenetic associations with disease has led to numerous methodological questions. As part of GAW20, our working group of 8 research groups focused on gene searching methods.

Results

Although the methods were varied, we identified 3 main themes within our group. First, many groups tackled the question of how best to use pedigree information in downstream analyses, finding that (a) the use of kinship matrices is common practice, (b) ascertainment corrections may be necessary, and (c) pedigree information may be useful for identifying parent-of-origin effects. Second, many groups also considered multimarker versus single-marker tests. Multimarker tests had modestly improved power versus single-marker methods on simulated data, and on real data identified additional associations that were not identified with single-marker methods, including identification of a gene with a strong biological interpretation. Finally, some of the groups explored methods to combine single-nucleotide polymorphism (SNP) and DNA methylation into a single association analysis.

Conclusions

A causal inference method showed promise at discovering new mechanisms of SNP activity; gene-based methods of summarizing SNP and DNA methylation data also showed promise. Even though numerous questions still remain in the analysis of DNA methylation data, our discussions at GAW20 suggest some emerging best practices.

     

A Mutual Self- and Informant-Report of Cognitive Complaint Correlates with Neuropathological Outcomes in Mild Cognitive Impairment

Background

This study examines whether different sources of cognitive complaint (i.e., self and informant) predict Alzheimer’s disease (AD) neuropathology in elders with mild cognitive impairment (MCI).

Methods

Data were drawn from the National Alzheimer’s Coordinating Center Uniform and Neuropathology Datasets (observational studies) for participants with a clinical diagnosis of MCI and postmortem examination (n = 1843, 74±8 years, 52% female). Cognitive complaint (0.9±0.5 years prior to autopsy) was classified into four mutually exclusive groups: no complaint, self-only, informant-only, or mutual (both self and informant) complaint. Postmortem neuropathological outcomes included amyloid plaques and neurofibrillary tangles. Proportional odds regression related complaint to neuropathology, adjusting for age, sex, race, education, depressed mood, cognition, APOE4 status, and last clinical visit to death interval.

Results

Mutual complaint related to increased likelihood of meeting NIA/Reagan Institute (OR = 6.58, p = 0.004) and Consortium to Establish a Registry for Alzheimer’s Disease criteria (OR = 5.82, p = 0.03), and increased neurofibrillary tangles (OR = 3.70, p = 0.03), neuritic plaques (OR = 3.52, p = 0.03), and diffuse plaques (OR = 4.35, p = 0.02). Informant-only and self-only complaint was not associated with any neuropathological outcome (all p-values>0.12).

Conclusions

In MCI, mutual cognitive complaint relates to AD pathology whereas self-only or informant-only complaint shows no relation to pathology. Findings support cognitive complaint as a marker of unhealthy brain aging and highlight the importance of obtaining informant corroboration to increase confidence of underlying pathological processes.     

Optical hyperspectral imaging in microscopy and spectroscopy – a review of data acquisition

Rather than simply acting as a photographic camera capturing two‐dimensional (x, y) intensity images or a spectrometer acquiring spectra (λ), a hyperspectral imager measures entire three‐dimensional (x, y, λ) datacubes for multivariate analysis, providing structural, molecular, and functional information about biological cells or tissue with unprecedented detail. Such data also gives clinical insights for disease diagnosis and treatment. We summarize the principles underpinning this technology, highlight its practical implementation, and discuss its recent applications at microscopic to macroscopic scales.

Datacube acquisition strategies in hyperspectral imaging x, y, spatial coordinates; λ, wavelength.     

Societal preferences for the return of incidental findings from clinical genomic sequencing: a discrete-choice experiment

Background:An important challenge with the application of next-generation sequencing technology is the possibility of uncovering incidental genomic findings. A paucity of evidence on personal utility for incidental findings has hindered clinical guidelines. Our objective was to estimate personal utility for complex information derived from incidental genomic findings.Methods:We used a discrete-choice experiment to evaluate participants’ personal utility for the following attributes: disease penetrance, disease treatability, disease severity, carrier status and cost. Study participants were drawn from the Canadian public. We analyzed the data with a mixed logit model.Results:In total, 1200 participants completed our questionnaire (available in English and French). Participants valued receiving information about high-penetrance disorders but expressed disutility for receiving information on low-penetrance disorders. The average willingness to pay was $445 (95% confidence interval [CI] $322–$567) to receive incidental findings in a scenario where clinicians returned information about high-penetrance, medically treatable disorders, but only 66% of participants (95% CI 63%–71%) indicated that they would choose to receive information in that scenario. On average, participants placed an important value ($725, 95% CI $600–$850) on having a choice about what type of findings they would receive, including receipt of information about high-penetrance, treatable disorders or receipt of information about high-penetrance disorders with or without available treatment. The predicted uptake of that scenario was 76% (95% CI 72%–79%).Interpretation:Most participants valued receiving incidental findings, but personal utility depended on the type of finding, and not all participants wanted to receive incidental results, regardless of the potential health implications. These results indicate that to maximize benefit, participant-level preferences should inform the decision about whether to return incidental findings.Clinical genomic sequencing technologies are on the verge of allowing individualized care at reasonable cost.¹ Patients and their families will soon receive information from clinical sequencing that has implications for clinical care, including information on consequences related to disease prognosis, treatment response or hereditary risk for disease.² Clinical sequencing can also generate incidental findings, which are clinically relevant genetic variants for disorders unrelated to the reason for ordering the genetic testing. The decision of whether to provide information about incidental findings is complex because such results will have varying clinical validity (whether the genetic variant causes the disorder) and utility (whether effective medical treatment is available for the disorder).^3,4 For example, although effective medical treatment may be available for some validated incidental findings, other incidental findings may not be validated as causing the disorder, and still others will be validated but not associated with effective treatment options.To address in part the challenges surrounding the return of incidental findings, the American College of Medical Genetics and Genomics published recommendations for reporting incidental findings from clinical sequencing.⁵ The statement lists a minimum of 56 genes that laboratories should examine, with results reported to patients through the managing physician. This list includes genes with high-penetrance mutations (i.e., a high proportion of individuals with the mutation will exhibit clinical symptoms) validated to be associated with disorders for which medical interventions are available.The original version of this statement did not “favour offering the patient a preference” for which results would be returned. The reasoning was that clinicians have a duty to prevent potential harm by telling patients about incidental findings. The working group that developed the recommendations further stated that it is impractical to provide the level of genetic counselling required for informed preference on all potential disorders.⁵ As such, the working group recommended that clinicians discuss with patients the possibility of receiving incidental findings from the list. It was argued that patient autonomy is preserved because patients can decline clinical sequencing if they prefer to not receive information about incidental findings.⁵ However, this rationale has been subject to debate because of its “all-or-none” nature, whereby patients must agree to receive information about incidental findings or clinical sequencing is not provided.^6–9 In April 2014, in response to the ongoing debate, the statement was amended to include an “opt-out” option for patients who do not want to receive information about incidental findings.¹⁰Notwithstanding the ethical debate, there is a lack of quantitative, preference-based economic evidence for the return of incidental genomic findings.⁸ It has been argued⁸ that this gap in evidence hindered development of the working group’s recommendation statement. More generally, evidence on preferences for the return of incidental findings is crucial for health policy, for health systems planning and for informing future lists that may include “many more genes.”⁸ We aimed to generate evidence on the personal utility that study participants from the Canadian public ascribe to the return of incidental genomic findings in the clinical setting. We chose participants from the general public because the public is the largest stakeholder in Canada’s publicly funded health care system.     

Quantification of liver fibrosis via second harmonic imaging of the Glisson's capsule from liver surface

Liver surface is covered by a collagenous layer called the Glisson's capsule. The structure of the Glisson's capsule is barely seen in the biopsy samples for histology assessment, thus the changes of the collagen network from the Glisson's capsule during the liver disease progression are not well studied. In this report, we investigated whether non‐linear optical imaging of the Glisson's capsule at liver surface would yield sufficient information to allow quantitative staging of liver fibrosis. In contrast to conventional tissue sections whereby tissues are cut perpendicular to the liver surface and interior information from the liver biopsy samples were used, we have established a capsule index based on significant parameters extracted from the second harmonic generation (SHG) microscopy images of capsule collagen from anterior surface of rat livers. Thioacetamide (TAA) induced liver fibrosis animal models was used in this study. The capsule index is capable of differentiating different fibrosis stages, with area under receiver operating characteristics curve (AUC) up to 0.91, making it possible to quantitatively stage liver fibrosis via liver surface imaging potentially with endomicroscopy.

     

Hydrolytic Activity and Substrate Specificity of an Endoglucanase from Zea mays Seedling Cell Walls

An endoglucanase was isolated from cell walls of Zea mays seedlings. Characterization of the hydrolytic activity of this glucanase using model substrates indicated a high specificity for molecules containing intramolecular (1→3),(1→4)-β-d-glucosyl sequences. Substrates with (1→4)-β-glucosyl linkages, such as carboxymethylcellulose and xyloglucan were, degraded to a limited extent by the enzyme, whereas (1→3)-β-glucans such as laminarin were not hydrolyzed. When (1→3),(1→4)-β-d-glucan from Avena endosperm was used as a model substrate a rapid decrease in vicosity was observed concomitant with the formation of a glucosyl polymer (molecular weight of 1-1.5 × 10⁴). Activity against a water soluble (1→3),(1→4)-β-d-glucan extracted from Zea seedling cell walls revealed the same depolymerization pattern. The size of the limit products would indicate that a unique recognition site exists at regular intervals within the (1→3),(1→4)-β-d-glucan molecule. Unique oligosaccharides isolated from the Zea (1→3),(1→4)-β-d-glucan that contained blocks of (1→4) linkages and/or more than a single contiguous (1→3) linkage were hydrolyzed by the endoglucanase. The unique regions of the (1→3),(1→4)-β-d-glucan may be the recognition-hydrolytic site of the Zea endoglucanase.     

Physico-chemical interpretation with cells electrophoresis of some antigen -- antibodies reactions in immuno-hematology.

The authors report the action of different antibodies of complete or incomplete type on the electrophoretic mobility of red-blood cells. They try to explain the results in terms of physicochemical stability. They make a distinction between three main possible classes concerning the mechanisms of action: a. direct diminution of the Zeta potential due to the antibody which in this case, is agglutinant. b. diminution of electrocinetic potential by "etching" of ionized groups responsible of the charge. To this mechanism, we could link the increase of the ionic force. c. Artificial increase of the dielectric constant by macromolecules.