Kenneth A. Marx

DNA visual and analytic data mining

Describes data exploration techniques designed to classify DNA sequences. Several visualization and data mining techniques were used to validate and attempt to discover new methods for distinguishing coding DNA sequences (exons) from non-coding DNA sequences (introns). The goal of the data mining was to see whether some other, possibly non-linear combination of the fundamental position-dependent DNA nucleotide frequency values could be a better predictor than the AMI (average mutual information). We tried many different classification techniques including rule-based classifiers and neural networks. We also used visualization of both the original data and the results of the data mining to help verify patterns and to understand the distinction between the different types of data and classifications. In particular, the visualization helped us develop refinements to neural network classifiers, which have accuracies as high as any known method. Finally, we discuss the interactions between visualization and data mining and suggest an integrated approach.

Statistical mechanical simulation of polymeric DNA melting with MELTSIM.

MOTIVATION MELTSIM is a windows-based statistical mechanical program for simulating melting curves of DNAs of known sequence and genomic dimensions under different conditions of ionic strength with great accuracy. The program is useful for mapping variations of base compositions of sequences, conducting studies of denaturation, establishing appropriate conditions for hybridization and renaturation, determinations of sequence complexity, and sequence divergence. RESULTS Good agreement is achieved between experimental and calculated melting curves of plasmid, bacterial, yeast and human DNAs. Denaturation maps that accompany the calculated curves indicate non-coding regions have a significantly lower (G+C) composition than coding regions in all species examined. Curves of partially sequenced human DNA suggest the current database may be heavily biased with coding regions, and excluding large (A+T)-rich elements. AVAILABILITY MELTSIM 1.0 is available at: //www.uml.edu/Dept/Chem/UMLBIC/Apps/MEL TSIM/MELTSIM-1.0-Win/meltsim. zip. Melting curve plots in this paper were made with GNUPLOT 3.5, available at: http://www.cs.dartmouth.edu/gnuplot_inf o.html Contact : blake@maine.maine.edu;

A quartz crystal microbalance cell biosensor: detection of microtubule alterations in living cells at nM nocodazole concentrations☆

The quartz crystal microbalance (QCM) was used to create a piezoelectric biosensor utilizing living endothelial cells (ECs) as the biological signal transduction element. ECs adhere to the hydrophilically treated gold QCM surface under growth media containing serum. At 24 h following cell addition, calibration curves were constructed relating the steady state Deltaf and DeltaR shift values observed to the numbers of electronically counted cells requiring trypsinization to be removed from the surface. We then utilized this EC QCM biosensor for the detection of the effect of [nocodazole] on the steady state Deltaf and DeltaR shift values. Nocodazole, a known microtubule binding drug, alters the cytoskeletal properties of living cells. At the doses used in these studies (0.11-15 microM), nocodazole, in a dose dependent fashion, causes the depolymerization of microtubules in living cells. This leads a monolayer of well spread ECs to gradually occupy a smaller area, lose cell to cell contact, exhibit actin stress fibers at the cell periphery and acquire a rounded cell shape. We observed the negative Deltaf shift values and the positive DeltaR shift values to increase significantly in magnitude over a 4-h incubation period following nocodazole addition, in a dose dependent fashion, with a transition midpoint of 900 nM. Fluorescence microscopy of the ECs, fixed on the gold QCM surface and stained for actin, demonstrated that the shape and cytoskeleton of ECs were affected by as little as 330 nM nocodazole. These results indicate that the EC QCM biosensor can be used for the study of EC attachment and to detect EC cytoskeletal alterations. We suggest the potential of this cellular biosensor for the real time identification or screening of all classes of biologically active drugs or biological macromolecules that affect cellular attachment, regardless of their molecular mechanism of action.

Quartz Crystal Microbalance Study of Endothelial Cell Number Dependent Differences in Initial Adhesion and Steady-State Behavior: Evidence for Cell-Cell Cooperativity in Initial Adhesion and Spreading

The quartz crystal microbalance (QCM) technique has been applied to the real time monitoring of endothelial cell (EC) adhesion and spreading on the QCM gold surface. We previously showed that the measured QCM Δf and ΔR shifts were due to cells adhering to the gold crystal surface, requiring proteolytic enzyme treatment to be removed from the surface, in order for the Δf and ΔR shifts to return to zero. In the present report, we demonstrate the quantitative dependence and saturation of the measured Δf and ΔR shifts on the number of firmly attached ECs as measured by electronic counting of the cells. We demonstrate through a light microscope simulation experiment that the different Δf and ΔR regions of the QCM temporal response curve correspond to the incident ECs contacting the surface, followed by their adhesion and spreading, which reflect cellular mass distribution and cytoskeletal viscoelasticity changes. Also, we demonstrate that the dose response curve of Δf and ΔR values versus attached EC number is more sensitive and possesses less scatter for the hydrophilically treated surface compared to the native gold surface of the QCM. For both surfaces, a Δf and ΔR versus trypsinized, attached EC number plot 1 h post‐seeding exhibits a sigmoid curve shape whereas a similar plot 24 h post‐seeding exhibits a hyperbolic curve shape. This number dependence suggests cell‐cell cooperativity in the initial cell adhesion and spreading processes. These QCM data and our interpretation are corroborated by differences in cell appearance and spreading behavior we observed for ECs in a light microscope fluorescence simulation experiment of the cell density effect. For a stably attached EC monolayer at 24 h post‐addition, steady‐state Δf and ΔR values are higher and exhibit saturation behavior for both the hydrophilically treated gold surface as compared to the untreated surface. The steady‐state 24 h Δf and ΔR values of stably attached ECs are shifted from the 1 h attached ECs. The 24 h values are characteristic of a more energy‐dissipative structure. This is consistent with the time‐dependent elaboration of surface contacts in anchorage‐dependent ECs via the attachment of intregrins to underlying extracellular matrix. It is also in agreement with the known energy dissipation function of the ECs that cover the interior of blood vessels and are exposed to continuous pulsatile blood flow.

Applications of Machine Learning and High‐Dimensional Visualization in Cancer Detection, Diagnosis, and Management

Abstract: Recent technical advances in combinatorial chemistry, genomics, and proteomics have made available large databases of biological and chemical information that have the potential to dramatically improve our understanding of cancer biology at the molecular level. Such an understanding of cancer biology could have a substantial impact on how we detect, diagnose, and manage cancer cases in the clinical setting. One of the biggest challenges facing clinical oncologists is how to extract clinically useful knowledge from the overwhelming amount of raw molecular data that are currently available. In this paper, we discuss how the exploratory data analysis techniques of machine learning and high‐dimensional visualization can be applied to extract clinically useful knowledge from a heterogeneous assortment of molecular data. After an introductory overview of machine learning and visualization techniques, we describe two proprietary algorithms (PURS and RadViz™) that we have found to be useful in the exploratory analysis of large biological data sets. We next illustrate, by way of three examples, the applicability of these techniques to cancer detection, diagnosis, and management using three very different types of molecular data. We first discuss the use of our exploratory analysis techniques on proteomic mass spectroscopy data for the detection of ovarian cancer. Next, we discuss the diagnostic use of these techniques on gene expression data to differentiate between squamous and adenocarcinoma of the lung. Finally, we illustrate the use of such techniques in selecting from a database of chemical compounds those most effective in managing patients with melanoma versus leukemia.

Vectorized Radviz and Its Application to Multiple Cluster Datasets

Radviz is a radial visualization with dimensions assigned to points called dimensional anchors (DAs) placed on the circumference of a circle. Records are assigned locations within the circle as a function of its relative attraction to each of the DAs. The DAs can be moved either interactively or algorithmically to reveal different meaningful patterns in the dataset. In this paper we describe Vectorized Radviz (VRV) which extends the number of dimensions through data flattening. We show how VRV increases the power of Radviz through these extra dimensions by enhancing the flexibility in the layout of the DAs. We apply VRV to the problem of analyzing the results of multiple clusterings of the same data set, called multiple cluster sets or cluster ensembles. We show how features of VRV help discern patterns across the multiple cluster sets. We use the Iris data set to explain VRV and a newt gene microarray data set used in studying limb regeneration to show its utility. We then discuss further applications of VRV.

Biochemical effects of binding [(H2O)(NH3)5RuII]2+ to DNA and oxidation to[(NH3)5RuIII]n—DNA.

Abstract The reaction of [(H 2 O)(NH 3 ) 5 Ru II ] 2+ with calf thymus and salmon sperm DNA has been studied over a wide fange of transition metal ion concentrations. Kinetic studies revealcd a biphasic reaction with an initial fairly rapid coordination of the metal ion being followed by slower reactions. Binding studies were done under pseudo-equilibrium conditions following completion of the initial rapid reaction. Spectra and HPLC of acid-hydrolyzed samples of [(NH 3 ) 5 Ru II ] n -DNA prepared by incubation of [(H 2 O)(NH 3 ) 5 Ru II ] 2+ with DNA (where [P DNA ] = 1.5 mM and reactant [Ru II ]/[P DNA ] ratios were in the fange 0.1 to 0.3) followed by air oxidation showed the predominant binding site on helical DNA to be in the major groove at the N-7 of guanine. The equilibrium constant for [(H 2 O)(NH 3 ) 5 Ru II ] 2+ binding to the G 7 site in helical CT DNA is 5.1 x 103. Differential pulse voltammetry exhibited a single peak at 48 mV, which is attributed to the reduction of Rum on the G 7 sites. At [Run]/[P DNA ] Tm values for the DNA decreased linearly with increasing ruthenium concentration and an increase in the intensity of the 565 nm dG→ Ru(III) charge transfer band was noted upon melting. The UV and CD spectra of these samples indicated no extensive destacking or alteration in geometry (B family) compared to unsubstituted DNA. At [Run]/[P DNA ]〉 0.5 or when single-stranded DNA was used, increased absorbance at 530 nm and 480 nm suggested additional binding to the exocyclic amine sites of adenine and cytosine residues. HPLC and individual spectrophotometric identification of the products derived from hydrolysis of these spec~es yielded both [(Gua)(NH 3 ) 5 Ru III ] and [(Ade)(NH 3 ) 5 Ru III ]. Earlier studies have established the cytidine and adenosine binding sites of [(NH 3 ) 5 Ru III ] to be at their exocyclic amines (C4 and A6). Coordination to these positions indicates disruption of the double helix since these amines are involved in hydrogen bonding on the interior of B-DNA. Agrose gel electrophoresis of superhelical pBR322 plasmid DNA after exposure to various complexes of [(nh 3 ) 5 Ru iii ] in the presence of a reductant and air generally revealed moderately efficient cleavage of the DNA, presumably due to the generation of hydroxyl radical via Fentons chemistry. However, similar studies involving [(NH 3 ) 5 Ru III ] directly coordinated to the DNA showed no strand cutting above background. Polyacrylamide gel electrophoresis of a 381 bp, 3′- 32 P-labeled fragment of pBR322 plasmid DNA containing low levels of bound [(NH 3 ) 5 Ru III ] further indicated negligible DNA cutting by the coordinated metal ion.

A gel electrophoresis study of the competitive effects of monovalent counterion on the extent of divalent counterions binding to DNA.

The behavior of alkaline earth metal cations (Mg2+ and Ca2+) and transition metal cations (Zn2+ and Cu2+) interacting with lambda-DNA-HindIII fragments ranging from 2,027 to 23,130 bp in Tris-borate-EDTA buffer solutions was investigated. The divalent counterions competed with Tris+ and Na+ for binding to polyion DNA, and the competition binding situations were investigated by measuring the reduction of the DNA mobility, by pulsed- or constant-field gel electrophoresis. The interaction of Mg2+ with DNA was intensively studied over a wide range of Mg2+ concentrations. In addition, we examined the competition binding as a function of ionic strength and DNA size. To compare valence effects, we studied Co(NH3)6(3+) interaction with DNA fragments under conditions similar to that of Mg2+. At relatively low Mg2+ concentration, the normalized titration curves of DNA mobility were well fit by Mannings two-variable counterion condensation (CC) theory. The agreement between the predicted value (total charge neutralization fraction theta) from Mannings CC theory and the data based on our measured DNA electrophoretic mobility reduction was consistent under our experimental conditions. In contrast to alkaline earth metal cations (Mg2+ and Ca2+), different binding behaviors were observed for the transition metal cations (Zn2+ and Cu2+). These differences highlight the usefulness of our reduced DNA electrophoretic mobility measurement approach to describing cation interactions with polyelectrolyte DNA.

TRIVALENT COUNTERION CONDENSATION ON DNA MEASURED BY PULSE GEL ELECTROPHORESIS

Pulse gel electrophoresis was used to measure the reduction of mobilities of lambda-DNA-Hind III fragments ranging from 23.130 to 2.027 kilobase pairs in Tris borate buffer solutions mixed with either hexammine cobalt(III), or spermidine3+ trivalent counterions that competed with Tris+ and Na+ for binding onto polyion DNA. The normalized titration curves of mobility were well fit by the two-variable counterion condensation theory. The agreement between measured charge fraction neutralized and counterion condensation prediction was good over a relatively wide range of trivalent cation concentrations at several solution conditions (pH, ionic strength). The effect of ionic strength, trivalent cation concentration, counterion structure, and DNA length on the binding were discussed based on the experimental measurements and the counterion condensation theory.

Studies of DNA Organization in Hydrated Spermidine-Condensed DNA Toruses and Spermidine-DNA Fibres

We have investigated in vitro spermidine-condensed DNA preparations by both biochemical and freeze-etch electron microscopic approaches. These studies lead us to the conclusion that the reversibly condensed preparations, qualitatively described by Mannings counterion condensation theory, contain disk-like torus structures largely comprised of unidirectional, circumferentially wrapped DNA. Stereoscopic measurements on stereomicrographs of DNA torus and fibre objects have demonstrated the feasibility of directly measuring DNA writhe or, for that matter, mapping any secondary, tertiary or quaternary structure features of a hydrated macromolecular array in which the features can be differentially highlighted by low replica metal shadow levels.