Lewis E. Lipkin

GELLAB: A computer system for 2D gel electrophoresis analysis I. Segmentation of spots and system preliminaries

Abstract GELLAB is a computerized analysis system for two-dimensional gel electrophoresis images. It is particularly concerned with tools for the automation of comparisons among multiple gels rather than simple gel-pair analyses. Interrogation of and experimentation with the spot data base may then be performed in order to extract measurements on particular spots within the set of gels. In the case of gels partitioned into multiple classes, it is possible to search for spots in the spot data base having statistically significant differences between classes. This first paper deals with: (1) some gel properties which bear on image acquisition and image analysis, (2) the broad image acquisition environment, beyond mere scanning, such as auxiliary data, standarization and calibration, and (3) segmentation as a means of spot extraction and characterization. Our segmentation algorithm is based on problem domain heuristics and particularly on properties of the second derivative. It is therefore largely shape and density independent. The segmenter, as the other GELLAB programs, is written in SAIL and runs on either a DECSYSTEM-10 or -20.

GELLAB: A computer system for 2D gel electrophoresis analysis. II. pairing spots

Abstract An algorithm and computer program CMPGEL, which is part of the GELLAB 2D electrophoretic gel analysis system, is described for pairing comparable spots in two gels. Pairing of spots between two gels at a time is necessary for the comparison of multiple gel images by the construction of multiple gel data bases. Interrogation of and experimentation with the spot data base may then be performed in order to extract measurements on particular spots for the set of gels. The enormous problem of pairing all spots in one gel with all spots in another gel is reduced to a small number of much simpler problems by partitioning spots in the gels into local regions. These are denoted by proximity to so called “landmark spots” and are termed “landmark regions.” Spots within landmark regions are then compared in order to perform the pairing.

Software aids for the analysis of 2D gel electrophoresis images.

Abstract A computer program, FLICKER, is described for aiding the analysis of two-dimensional gel electrophoresis images. It facilitates the comparison of parts of two or more images by flicker comparison as well as enabling the extraction and measurement of operator-selected regions. Measures of image distortion and the comparison of images are available for testing the validity of biochemical differences between the gels. The measurements available include transforming the density feature into a measure of polypeptide concentration.

Generating non-overlapping displays of nucleic acid secondary structure

A new algorithm is presented which permits the display of nucleic acid secondary structure by computer. This algorithm circumvents the problem of overlapping portions of the molecule which is inherent in some other drawing programs. The results from this algorithm may also be used as input to the drawing algorithm previously reported in this journal [1] to untangle most of a drawing. The algorithm also represents the molecule in a form which makes visual comparisons for similarity quite easy since it guarantees that comparable features will reside in the same relative position in the drawings when the drawings are normalized.

GELLAB: A computer system for two-dimensional gel electrophoresis analysis. III. Multiple two-dimensional gel analysis

Abstract This is the last of a series of three papers describing a computer analysis system, GELLAB, for aiding the digitization, segmentation, interimage comparisons, and analysis of two-dimensional (2D) gel electrophoresis images. The system is directed to the identification of biologically significant changes manifest as spot differences despite complications introduced by a wide variety of gel preparative, staining, detection, and digitization variances. The operations applicable to single-gel images (segmentation, etc.) are described in the first paper. The second deals with several of the elements that will appear in the fully developed data base; i.e., constructs developed from images taken two at a time, based on regional correspondences established by landmarks. This paper focuses on methods for handling multiple gels. The keystone of this analysis is the R-gel, the representative gel image. This acts as the major framework to link spots across the set of gels and serves as a practical substitute for an unattainable canonical model. It is the referent around which the total spot data base is organized. The primary data base consists of the totality of the lists of corresponding spots (R-spot sets) and their associated properties and interrelations. Interrogation of, and experimentation with, the data base allows the user to find and extract measurements of biological significance from these congeneric polypeptides. Some of the problems of and some solutions for dealing with a large number of possibly incomplete biologically determined intergel spot density distributions are presented. Solution strategies include constructing mosaics, using the digitized images of areas surrounding selected spots, and creating labeled map images of R-gel correspondences. Listings of relative position and density information of ordered R-spot sets is also of value in this interactive system which permits refining of the initial data base. The multiple-gel data base system is very general in that gel segmentation and spot-pairing algorithms other than those now used in GELLAB may be substituted in the early phases of data base construction. Discussion of CGEL, the GELLAB program for constructing, partitioning, searching, retrieving, and formatting the data base, is illustrated using results obtained from 2D gel data of phytohemagglutinin (PHA) stimulation of lymphocytes. Experience with utilizing GELLAB on a variety of biological problems has suggested potential system developments and new system features which are noted. GELLAB runs on a DECSYSTEM-20 (or DECSYSTEM-10) and was written in SAIL.

AUTOMATED AUTORADIOGRAPHIC GRAIN COUNTING IN HUMAN DETERMINED CONTEXT

An automated autoradiographic grain counter which dichotomizes results into nuclear and background categories for multiple human determined classes of Romanowsky-stained cells has been constructed. Additional instrument capabilities and functions are indicated as applicable to appropriately prepared material. The operation of the instrument is summarized. Its performance characteristics and its potential role in general quantitative cytology, cytokinetics research and cancer chemotherapy will be evaluated by projected field trial instruments.

A REAL TIME PICTURE PROCESSOR FOR USE IN BIOLOGIC CELL IDENTIFICATION II. HARDWARE IMPLEMENTATION

Hardware design specifications for a real time picture processor are presented. Attention is concentrated on those unique aspects of the hardware that make possible high speed storage and processing of images. The latter are the major contributors to rapidity by which various image transforms may be accomplished—speeds which are great enough to warrant the term real time.

A REAL TIME PICTURE PROCESSOR FOR USE IN BIOLOGIC CELL IDENTIFICATION I. SYSTEM DESIGN

We show that the class of special purpose picture processors has common characteristics which set it apart from conventional general purpose computers. We argue for the implementation of a fast serial processor of gray scale images having easy access to a tessellated gray scale neighborhood, where the tessellation of a neighborhood is the movement of that neighborhood through the image array in some deterministic way. Such a machine would be very useful in an interactive system used for problems such as the design of effective procedures for automated cell classification.

HIGH RESOLUTION SHADING CORRECTION

Quantitative cytology requires accurate representation of a specimens optical densities. As the requirements for measurement precision increase, instrument-induced errors become increasingly more difficult to reduce to the point at which their effect on experimental data is insignificant compared to the measured parameters. Shading induces a significant amount of amplitude ambiguity to data obtained from a scanning system. A method of shading correction on single pixels is introduced as a new way to reduce some errors that currently plague scanning systems.

DNAase I hypersensitive sites may be correlated with genomic regions of large structural variation

Helical-twist, roll and torsion-angle variations calculated by the Calladine (1982)-Dickerson (1983) rules were scanned along several nucleotide sequences for which DNAase I cleavage data are available. It has been shown that for short synthetic oligomers DNAase I cuts preferentially at positions of high helical twist (Dickerson & Drew, 1981; Lomonossoff et al., 1981). Our calculations indicate that DNAase I sensitive and hypersensitive sites in chromatin are correlated with regions of successive, large, helical-twist angle variations from regular B-DNA. In many cases these regions exhibit large variations in base-pair roll and backbone torsion angles as well. It has been suggested that DNAase I cuts in the vicinity of cruciforms. However, it was recently demonstrated by Courey & Wang (1983) and Gellert et al. (1983) that such cruciform formation in a negatively supercoiled DNA is kinetically forbidden under physiological conditions. We thus propose that clustering of large twist-angle (and/or roll and backbone torsion angle) variations may be among the conformational features recognized by the enzyme. Specific cuts can then preferentially occur at base-pair steps with high helical twists.