Amir H. Assadi

Probabilistic interaction network of evidence algorithm and its application to complete labeling of peak lists from protein NMR spectroscopy.

The process of assigning a finite set of tags or labels to a collection of observations, subject to side conditions, is notable for its computational complexity. This labeling paradigm is of theoretical and practical relevance to a wide range of biological applications, including the analysis of data from DNA microarrays, metabolomics experiments, and biomolecular nuclear magnetic resonance (NMR) spectroscopy. We present a novel algorithm, called Probabilistic Interaction Network of Evidence (PINE), that achieves robust, unsupervised probabilistic labeling of data. The computational core of PINE uses estimates of evidence derived from empirical distributions of previously observed data, along with consistency measures, to drive a fictitious system M with Hamiltonian H to a quasi-stationary state that produces probabilistic label assignments for relevant subsets of the data. We demonstrate the successful application of PINE to a key task in protein NMR spectroscopy: that of converting peak lists extracted from various NMR experiments into assignments associated with probabilities for their correctness. This application, called PINE-NMR, is available from a freely accessible computer server (http://pine.nmrfam.wisc.edu). The PINE-NMR server accepts as input the sequence of the protein plus user-specified combinations of data corresponding to an extensive list of NMR experiments; it provides as output a probabilistic assignment of NMR signals (chemical shifts) to sequence-specific backbone and aliphatic side chain atoms plus a probabilistic determination of the protein secondary structure. PINE-NMR can accommodate prior information about assignments or stable isotope labeling schemes. As part of the analysis, PINE-NMR identifies, verifies, and rectifies problems related to chemical shift referencing or erroneous input data. PINE-NMR achieves robust and consistent results that have been shown to be effective in subsequent steps of NMR structure determination.

HYPOTrace: Image Analysis Software for Measuring Hypocotyl Growth and Shape Demonstrated on Arabidopsis Seedlings Undergoing Photomorphogenesis

Analysis of time series of images can quantify plant growth and development, including the effects of genetic mutations (phenotypes) that give information about gene function. Here is demonstrated a software application named HYPOTrace that automatically extracts growth and shape information from electronic gray-scale images of Arabidopsis (Arabidopsis thaliana) seedlings. Key to the method is the iterative application of adaptive local principal components analysis to extract a set of ordered midline points (medial axis) from images of the seedling hypocotyl. Pixel intensity is weighted to avoid the medial axis being diverted by the cotyledons in areas where the two come in contact. An intensity feature useful for terminating the midline at the hypocotyl apex was isolated in each image by subtracting the baseline with a robust local regression algorithm. Applying the algorithm to time series of images of Arabidopsis seedlings responding to light resulted in automatic quantification of hypocotyl growth rate, apical hook opening, and phototropic bending with high spatiotemporal resolution. These functions are demonstrated here on wild-type, cryptochrome1, and phototropin1 seedlings for the purpose of showing that HYPOTrace generated expected results and to show how much richer the machine-vision description is compared to methods more typical in plant biology. HYPOTrace is expected to benefit seedling development research, particularly in the photomorphogenesis field, by replacing many tedious, error-prone manual measurements with a precise, largely automated computational tool.

Detection of a gravitropism phenotype in glutamate receptor-like 3.3 mutants of Arabidopsis thaliana using machine vision and computation.

Gene disruption frequently produces no phenotype in the model plant Arabidopsis thaliana, complicating studies of gene function. Functional redundancy between gene family members is one common explanation but inadequate detection methods could also be responsible. Here, newly developed methods for automated capture and processing of time series of images, followed by computational analysis employing modified linear discriminant analysis (LDA) and wavelet-based differentiation, were employed in a study of mutants lacking the Glutamate Receptor-Like 3.3 gene. Root gravitropism was selected as the process to study with high spatiotemporal resolution because the ligand-gated Ca2+-permeable channel encoded by GLR3.3 may contribute to the ion fluxes associated with gravity signal transduction in roots. Time series of root tip angles were collected from wild type and two different glr3.3 mutants across a grid of seed-size and seedling-age conditions previously found to be important to gravitropism. Statistical tests of average responses detected no significant difference between populations, but LDA separated both mutant alleles from the wild type. After projecting the data onto LDA solution vectors, glr3.3 mutants displayed greater population variance than the wild type in all four conditions. In three conditions the projection means also differed significantly between mutant and wild type. Wavelet analysis of the raw response curves showed that the LDA-detected phenotypes related to an early deceleration and subsequent slower-bending phase in glr3.3 mutants. These statistically significant, heritable, computation-based phenotypes generated insight into functions of GLR3.3 in gravitropism. The methods could be generally applicable to the study of phenotypes and therefore gene function.

Classification of patients with pain based on neuropathic pain symptoms: Comparison of an artificial neural network against an established scoring system

Wider use of pain assessment tools that are specifically designed for certain types of pain – such as neuropathic pain – contribute an increasing amount of information which in turn offers the opportunity to employ advanced methods of data analysis. In this manuscript, we present the results of a study where we employed artificial neural networks (ANNs) in an analysis of pain descriptors with the goal of determining how an approach that uses a specific symptoms‐based tool would perform with data from the real world of clinical practice. We also used traditional statistics approaches in the form of established scoring systems as well as logistic regression analysis for the purpose of comparison. Our results confirm the clinical experience that groups of pain descriptors rather than single items differentiate between patients with neuropathic and non‐neuropathic pain. The accuracy obtained by ANN analysis was only slightly higher than that of the traditional approaches, indicating the absence of nonlinear relationships in this dataset. Data analysis with ANNs provides a framework that extends what current approaches offer, especially for dynamic data, such as the rating of pain descriptors over time.

Unsupervised clustering algorithm for N-dimensional data

Cluster analysis is an important tool for classifying data. Established techniques include k-means and k-median cluster analysis. However, these methods require the user to provide a priori estimations of the number of clusters and their approximate location in the parameter space. Often these estimations can be made based on some prior understanding about the nature of the data. Alternatively, the user makes these estimations based on visualization of the data. However, the latter is problematic in data sets with large numbers of dimensions. Presented here is an algorithm that can automatically provide these estimates without human intervention based on the inherent structure of the data set. The number of dimensions does not limit it.

Dynamic PCA for network feature extraction in multi-electrode recording of neurophysiological data in cortical substrate of pain

Abstract A novel multi-scale analysis of multi-electrode spike recording during heat pain stimulation in rats is applied to quantify non-stationary patterns of neuronal response. This approach would allow biological constraints to translate into multi-dimensional geometry. We then determine the optimal scale, resolution and density of a neuronal localization that best characterizes the cortical response. Within the optimal choices, we determine the inherent dimension of the locally linear principal component analysis (PCA) that approximates the non-linear geometric structure of data, and minimizes the reconstruction error within the prescribed bounds. When dimension is one, two, or three, our optimization algorithms determine the system of non-linear principal curves that best approximates the data.

Nonlinear methods for clustering and reduction of dimensionality

Analysis of data in computational finance and computational neuroscience share a number of common traits: data are typically massive, noisy, very high dimensional, and governed by complete multi-scale time dynamics. The set of known parameters forms a small subset of the true variates that control the dynamics of the systems from which data is collected. Reduction of dimensionality of the data, and clustering of system parameters according to a relevant measure of independence, and improving signal to noise ratio, are among the core problems of both disciplines. We propose a nonlinear version of independent component analysis for clustering of parameters and separating clusters according to their measure of statistical independence. Analogously, we propose a nonlinear version of principal component analysis for reducing the dimensionality of data. The combination of these two methods forms the basis for a dynamic pattern recognition paradigm. This approach is inspired by a mathematical analogy to a successful method for estimation of patterns of functional connectivity in neuro-imaging.

Learning Gestalt of surfaces in natural scenes

We develop a computational model for scenes with surfaces that have rough and non-smooth small-scale structure but with a perceived global (larger-scale) geometric form. Examples include grass and meadow, surfaces textured with sand-paper, natural scenes having rough texture such as the skin of crocodile, pine cones, a field of sea urchins, forests, ripples and waves on water surfaces, etc. Another domain of examples arise in scientific exploration of microscopic images, such as the atomic force microscopy (AFM) images from alloys in materials science, molecular beam epitaxy (MBE), rough surfaces due to ballistic deposition (ED surfaces) and random deposition surfaces (RD). As a last example, one may translate some outstanding image processing problems of infra-red astronomy to understanding the random texture of clouds combined with noise, e.g. to describe algorithms that detect stars within noisy data provided by infra-red imaging devices.

Integral representations of finite transformation groups III: simply-connected four-manifolds

The action of a finite group G on a simply-connected 4-manifold X yields a ZG-lattice H2(X). In this paper we study the relations between the geometry of the G-action on X and the homological and representation theoretic properties of H2(X).

An application of support vector machines and symmetry to computational modeling of perception through visual attention

Abstract Eye movement is connected with attention and visual perception. Our previous research provided a computational model for detection of symmetry, and a case was made for a dynamic model of symmetry detection based on adaptive saccades and visual attention. Here, we present a computational model of saccade target selection and simulate its action in the context of perception of global periodic symmetry of surfaces using local (foveal) symmetry approximations to direct saccadic eye movements. Target selection is modeled via support vector machine regression. The motivation for support vector model finds its justification in the properties of the superior colliculus.