Technique of Gene Expression Profiles Extraction Based on the Complex Use of Clustering and Classification Methods.

In this paper, we present the results of the research concerning extraction of informative gene expression profiles from high-dimensional array of gene expressions considering the state of patients' health using clustering method, ML-based binary classifiers and fuzzy inference system. Applying of the proposed stepwise procedure can allow us to extract the most informative genes taking into account both the subtypes of disease or state of the patient's health for further reconstruction of gene regulatory networks based on the allocated genes and following simulation of the reconstructed models. We used the publicly available gene expressions data as the experimental ones which were obtained using DNA microarray experiments and contained two types of patients' gene expression profiles-the patients with lung cancer tumor and healthy patients. The stepwise procedure of the data processing assumes the following steps-in the beginning, we reduce the number of genes by removing non-informative genes in terms of statistical criteria and Shannon entropy; then, we perform the stepwise hierarchical clustering of gene expression profiles at hierarchical levels from 1 to 10 using the SOTA (Self-Organizing Tree Algorithm) clustering algorithm with correlation distance metric. The quality of the obtained clustering was evaluated using the complex clustering quality criterion which is considered both the gene expression profiles distribution relative to center of the clusters where these gene expression profiles are allocated and the centers of the clusters distribution. The result of this stage execution was a selection of the optimal cluster at each of the hierarchical levels which corresponded to the minimum value of the quality criterion. At the next step, we have implemented a classification procedure of the examined objects using four well known binary classifiers-logistic regression, support-vector machine, decision trees and random forest classifier. The effectiveness of the appropriate technique was evaluated based on the use of ROC (Receiver Operating Characteristic) analysis using criteria, included as the components, the errors of both the first and the second kinds. The final decision concerning the extraction of the most informative subset of gene expression profiles was taken based on the use of the fuzzy inference system, the inputs of which are the results of the appropriate single classifiers operation and the output is the final solution concerning state of the patient's health. To our mind, the implementation of the proposed stepwise procedure of the informative gene expression profiles extraction create the conditions for the increasing effectiveness of the further procedure of gene regulatory networks reconstruction and the following simulation of the reconstructed models considering the subtypes of the disease and/or state of the patient's health.

A general method for determining molecular interfaces and layers.

A general and direct computational scheme to locate the surface separating arbitrarily shaped domains made up of molecules (or any other particles) has been developed and is described and illustrated for several, both artificial and physical examples. The proposed scheme consists of two modules: (i) triangulation and (ii) assignment of simplices to domains. Three different triangulation methods are employed, viz., the Delaunay triangulation, regular triangulation, and quasi-triangulation. In the triangulated system, the assignment step is carried out in two different ways, one based on the characteristic metric of a particular triangulation procedure and the other on the concept of a touching sphere. Some of the combinations of the triangulation and assignment steps lead to methods already used by others to find interfacial or surface molecules, namely the alpha-shape-based method of Usabiaga nad Duque [Phys. Rev. E 79 (2009) 046709] and GITIM of Sega et al. [J. Chem. Phys. 138 (2013) 044110]. The resulting surface is defined not only as a discrete set of particles, but it is build up of facets of the triangulation forming a broken line in two dimensions or a polyhedral surface in three dimensions. Individual molecular layers are identified in a very straightforward manner, starting with the interfacial layer itself and proceeding into the interior of the phase. The proposed scheme is illustrated first by identifying border molecules of pre-sampled domains of several shapes in a plane and then applied to five physically meaningful examples: thin films, near critical water, liquid water slab in an electric field, liquid water at a solid wall, and water at condition of electric-field-induced jetting. Performance of the considered methods is critically assessed. Treatment of domains forming percolating clusters through periodic boundary conditions is also described along with the determination of their periodicity and dimensionality.

Molecular simulation of aqueous electrolyte solubility. 2. Osmotic ensemble Monte Carlo methodology for free energy and solubility calculations and application to NaCl.

We present a new and computationally efficient methodology using osmotic ensemble Monte Carlo (OEMC) simulation to calculate chemical potential-concentration curves and the solubility of aqueous electrolytes. The method avoids calculations for the solid phase, incorporating readily available data from thermochemical tables that are based on well-defined reference states. It performs simulations of the aqueous solution at a fixed number of water molecules, pressure, temperature, and specified overall electrolyte chemical potential. Insertion/deletion of ions to/from the system is implemented using fractional ions, which are coupled to the system via a coupling parameter λ that varies between 0 (no interaction between the fractional ions and the other particles in the system) and 1 (full interaction between the fractional ions and the other particles of the system). Transitions between λ-states are accepted with a probability following from the osmotic ensemble partition function. Biasing weights associated with the λ-states are used in order to efficiently realize transitions between them; these are determined by means of the Wang-Landau method. We also propose a novel scaling procedure for λ, which can be used for both nonpolarizable and polarizable models of aqueous electrolyte systems. The approach is readily extended to involve other solvents, multiple electrolytes, and species complexation reactions. The method is illustrated for NaCl, using SPC/E water and several force field models for NaCl from the literature, and the results are compared with experiment at ambient conditions. Good agreement is obtained for the chemical potential-concentration curve and the solubility prediction is reasonable. Future improvements to the predictions will require improved force field models.

Hypertrophic cardiomyopathy: from mutation to functional analysis of defective protein.

AIM: To analyze the genesis of hypertrophic cardiomyopathy on a large cohort of patients from molecular genetics point of view and perform the functional analysis of the 3D molecular model of defective myosin-7 protein in silico. METHODS: The study enrolled 153 patients with diagnosed hypertrophic cardiomyopathy from different parts of the Czech Republic. DNA samples were analyzed for mutations in exons 21 and 22 of the MYH7 gene, which have been associated with high mutation clustering. The 3D model of human myosin-7 was built using the x-ray structure of nucleotide-free scallop myosin S1 as the structural template. We performed de novo structure prediction of mutant and wild type peptides spanning the 769-788 amino acids region of the myosin-7 protein. RESULTS: The Arg870His and Asp778Val amino acid alterations were found in 2 unrelated patients with a severe form of hypertrophic cardiomyopathy. The Asp778Val variation was chosen for subsequent 3D molecular modeling in silico. The mutation of the Asp by Val not only changes the character of the interaction pattern with other amino acids or ions but Val, being a small hydrophobic amino acid, can also completely change the stability of the region. CONCLUSION: Mutation location in the MYH7 gene and changes in amino acid composition may have a crucial negative impact on the outcome of the disease in patients with hypertrophic cardiomyopathy. In addition, a mutation that changes the charge of the amino acid is more likely to affect protein function than a conservative mutation.

Percolation threshold parameters of fluids.

Extensive Monte Carlo simulations on three qualitatively different model supercritical fluids (square-well fluid, Lennard-Jonesium, and primitive water) have been performed to examine percolation threshold parameters for continuum (correlated) models and their relation to general results valid for random lattice models; random-site percolation simple-cubic lattice has therefore been considered as well. Two different bond criteria, the configurational and self-bound ones, defining a cluster have been used. In addition to the percolation threshold occupation probability pc and the percolation threshold fluid density rhoc, the correlation length exponent nu and the wrapping probability at the percolation threshold Rw,c have also been evaluated. It is found that parameters nu and Rw,c exhibit not only strong temperature dependence but also, unlike the case of lattice systems, dependence on the nature of the system considered and the employed definition of the cluster.

Antitumor effects and cytotoxicity of recombinant plant nucleases.

Recombinant plant nucleases R-TBN1 and R-HBN1 were isolated to homogeneity and examined for their antitumor effects and cytotoxicity. Although antiproliferative effects of both recombinant nucleases were not significant on the ML-2 cell culture in vitro, the nucleases were strongly cytostatic in vivo after their administration intravenously as stabilized conjugates with polyethylene glycol (PEG). Recombinant nucleases were as effective against melanoma tumors as previously studied pine pollen (PN) and mung bean nucleases and their effects were reached at about 10 times lower concentrations compared to the use of bovine seminal RNase (BS-RNase). Because the recombinant nucleases R-HBN1 and R-TBN1 share only 67.4% amino acid identity and showed only partial immunochemical cross-reactivity, their similar anticancerogenic effects can be mainly explained by their catalytical similarity. Both recombinant nucleases showed lower degree of aspermatogenesis compared to BS-RNAse and PN nuclease. Unlike BS-RNase, aspermatogenesis induced by both recombinant nucleases could not be prevented by the homologous antibody complexes. Owing to relatively low cytotoxicity on the one hand, and high efficiency at low protein levels on the other, recombinant plant nucleases R-HBN1 and R-TBN1 appear to be stable biochemical agents that can be targeted as potential antitumor cytostatics.

Percolation transition in fluids: scaling behavior of the spanning probability functions.

We examine different spanning probability functions (wrapping and crossing) near the percolation threshold of a supercritical square-well fluid and determine the threshold values of these probabilities, which may be universal for all fluids. It is shown that for a continuous system, over a wide range of system size, the wrapping probabilities can be described by universal scaling functions, whereas the crossing probabilities do not show such universal behavior over the same range of system size. The obtained universal functions for the wrapping probabilities can be used for an estimation of the percolation threshold in fluids in general. The results for the crossing probabilities allow us then to characterize large clusters in real fluids.

Effect of wheat leaf ribonuclease on tumor cells and tissues.

The antiproliferative and antitumor effect of wheat leaf ribonuclease was tested in vitro on the human ML-2 cell line and in vivo on athymic nude mice bearing human melanoma tumors. The antiproliferative activity of this plant ribonuclease was negligible in comparison with bovine seminal ribonuclease. In the experiments in vivo, a significant decrease of the tumor size, however, was observed in the mice treated with wheat leaf ribonuclease (27 kDa) compared with the control RNase A and polyethylene glycol. In nude mice injected intratumoraly with wheat leaf ribonuclease, the tumor size decreased from 100% in the control mice to 39% in treated mice. In the mice treated with polyethylene glycol-conjugated wheat leaf ribonuclease, the tumor reduction was observed from 100 to 28%, whereas in counterparts treated with polyethylene glycol-conjugated bovine seminal ribonuclease the tumor inhibition was reduced from 100 to 33%. Certain aspermatogenic and embryotoxic activity of wheat leaf ribonuclease and bovine seminal ribonuclease also appeared, but was lower in comparison with the effect of onconase. Mutual immunological cross-reactivity between wheat leaf ribonuclease antigens on one side and animal RNases (bovine seminal ribonuclease, RNase A, human HP-RNase and onconase) on the other side proved a certain structural similarity between animal and plant ribonucleases. Immunogenicity of wheat leaf ribonuclease was weaker in comparison with bovine seminal ribonuclease (titer of antibodies 160-320 against 1280-2560 in bovine seminal ribonuclease). Interestingly, immunosuppressive effect of wheat leaf ribonuclease tested on mixed lymphocyte culture-stimulated human lymphocytes reached the same level as that of bovine seminal RNase. The antibodies against wheat leaf ribonuclease produced in the injected mice did not inactivate the biological effect of this plant RNase in vivo. This is probably the first paper in which plant ribonuclease was used as antiproliferative and antitumor drug against animal and human normal and tumor cells and tissues in comparison with animal ribonucleases.

Effect of hyaluronidase and PEG chain conjugation on the biologic and antitumor activity of RNase A.

Subcutaneous application of bovine RNase A conjugated to HYase (bovine hyaluronidase), polyethylene glycol (PEG) and HYase+PEG resulted in a marked reduction of the width of the spermatogenic layers of the mouse testes. The number of sperms in caput epididymidis was significantly decreased in mice injected with conjugated RNase A. There was not any significant embryotoxic effect of free RNase A even conjugated with HYse, PEG and HYse+PEG. The immunogenicity, expressed in production of antibodies against free RNase A or conjugates with PEG, was very low. However, the immunogenic action of this enzyme conjugated only to HYase was much higher and produced the same immunogenicity as HYase itself. The immunogenic effect of RNase A+HYase conjugate decreased when PEG was joined to this conjugate. The inhibitory effect of RNase A conjugated to HYase, PEG and HYase+PEG on human ML-2 cells studied in vitro, was practically ineffective. On the other side, when RNase A conjugated to HYase or PEG was administered intraperitoneally into the mice bearing human melanoma, the antitumor effect was pronounced.

PEG chains increase aspermatogenic and antitumor activity of RNase A and BS-RNase enzymes.

RNase A (bovine pancreatic ribonuclease) and BS-RNase (bovine seminal ribonuclease) are monomeric and dimeric enzymes, respectively, with aspermatogenic and antitumor activities. While the aspermatogenic and, in some experimental situations, the antitumor effects of the RNase A are only minor, the activity of BS-RNase in these phenomena is very significant. These differences can be annulled by means of conjugation of the enzymes with PEG (polyethylene glycol) chains. Aspermatogenic activity was studied histologically following subcutaneous injections of RNase A and BS-RNase conjugates in ICR mice, and the antitumor activity in athymic nude mice with growing human melanoma with i.p. injection of these conjugated ribonucleases. The experiments proved that RNase A, when conjugated to PEG, produced identical aspermatogenic and antitumour effects as BS-RNase conjugated to this polymer. Immunogenicity of RNase A and BS-RNase did not change substantially after the conjugation with PEG polymers. Binding of produced antibodies to both ribonucleases attached to PEG, however, was substantially reduced.