Learning and discovery from a clinical database: an incremental concept formation approach.

The main interest of this research is to discover clinical implications from a large PTCA (Percutaneous Transluminal Coronary Angioplasty) database. A case-based concept formation model D-UNIMEM, a modified version of Lebowitz's UNIMEM, is proposed for this purpose. In this model, we integrated two kinds of class memberships: the feature-disjunction class membership and the index-conjunction class membership. The former is a polythetic clustering approach and serves at the early stage of concept formation. The latter allows only relevant instances to be placed in the same cluster and serves as the later stage of concept formation. D-UNIMEM could extract interesting correlations among features from the learned concept hierarchy.

Enhanced synthesis of tumor necrosis factor-inducible proteins, plasminogen activator inhibitor-2, manganese superoxide dismutase, and protein 28/5.6, is selectively triggered by the 55-kDa tumor necrosis factor receptor in human melanoma cells.

We have demonstrated that A375 melanoma cells express mRNA for both types of tumor necrosis factor (TNF) receptors and receptor proteins on their plasma membranes. Specific agonist and blocking antibodies to either 55-kDa (TNF-R1) or 75-kDa (TNF-R2) TNF receptors combined with two-dimensional gel analysis were employed to determine which receptor type is responsible for mediating the induction of individual melanoma proteins. Our results indicate that the enhanced synthesis of proteins 21/>7 (M(r)/pI), 28/5.6, and 41/5.7 is selectively induced through TNF-R1. TNF induces these proteins; antagonist antibody to TNF-R1 prevents their induction by TNF, and TNF-R1 agonist induces them in the absence of TNF. Identification of these proteins by immunoblot analysis proved that 21/>7 is manganese superoxide dismutase, protein 28/5.6 is unrelated to 27/28-kDa heat shock protein, and protein 41/5.7 is plasminogen activator inhibitor-2. Furthermore, TNF cytotoxicity for A375 cells is also mediated by TNF-R1. These studies indicate that TNF-R1 is a critical signaling receptor for TNF action on A375 cells and demonstrate the potential use of TNF-R1 antibodies to selectively block or enhance specific effects of TNF on melanoma cells.

Characterization of TNF receptors on human thymocytes.

Although tumor necrosis factor-alpha (TNF) is constitutively expressed in human and mouse thymus, the effects of TNF on thymocyte proliferation, differentiation and survival suggest that its influence in the thymus is complex. To determine if this complexity results from changes in the expression of the two TNF receptors during thymocyte differentiation, we examined the expression of the 55 kDa TNF receptor (TNF-R1) and the 75 kDa TNF receptor (TNF-R2) on postnatal human thymocytes. Both TNF-R1 and TNF-R2 mRNA were found in resting human thymocytes by reverse transcriptase-polymerase chain reaction (RT-PCR). Using mAb which specifically react with the respective TNF receptors and a highly sensitive, three-step method of immunofluorescence, cell surface TNF-R1 was detected on the vast majority of thymocytes. In contrast, detectable cell surface TNF-R2 was present on a mean of only 12.9% of thymocytes. TNF conjugated to phycoerythrin (TNF-PE) also reacted with a small population of thymocytes and was found to specifically block binding of the TNF-R2 mAb and not the TNF-R1 mAb, implicating preferential binding of TNF-PE to TNF-R2. Using dual-color immunofluorescence with TNF-PE we found that the population of cells which express TNF-R2 also express high levels of the TCR alpha, beta-CD3 complex, CD4 or CD8, and IL-2 receptor alpha chain. Thus, immature (TCRneg/low) thymocytes express TNF-R1 while mature (TCRhigh) thymocytes can also express TNF-R2. This differential expression of TNF receptors provides a mechanism for distinct effects of TNF on immature vs. mature thymocytes.

Mass spectrometric and Edman sequencing of lipocortin I isolated by two-dimensional SDS/PAGE of human melanoma lysates.

We have integrated preparative two-dimensional polyacrylamide gel electrophoresis with high-performance tandem mass spectrometry and Edman degradation. By using this approach, we have isolated and identified, by partial sequencing, a human melanoma protein (34 kDa, pI 6.4) as lipocortin I. To our knowledge, this protein was not previously known to be associated with melanoma cells. The identity of the protein was confirmed by two-dimensional immunoblot analysis. High-energy collision-induced dissociation analysis revealed the sequence and acetylation of the N-terminal tryptic peptide and an acrylamide-modified cysteine in another tryptic peptide. Thus, knowledge concerning both the primary structure and covalent modifications of proteins isolated from two-dimensional gels can be obtained directly by this approach, which is applicable to a broad range of biological problems.

RUBIDIUM, a program for computer-aided assignment of two-dimensional NMR spectra of polypeptides.

Taking advantage of the rule-based expert system technology, a program named RUBIDIUM (Rule-Based Identification In 2D NMR Spectrum) was developed to accomplish the automatic 1H NMR resonance assignments of polypeptides. Besides noise elimination and peak selection capabilities, RUBIDIUM detects the cross-peak patterns of amino acid residues in the COSY spectrum, assigning these patterns to amino acid types, performing sequential assignments using combined COSY/NOESY spectra, and finally, achieving the total assignment of the 1H NMR spectrum.

An artificial-intelligence technique for qualitatively deriving enzyme kinetic mechanisms from initial-velocity measurements and its application to hexokinase.

We have developed a computer method based on artificial-intelligence techniques for qualitatively analysing steady-state initial-velocity enzyme kinetic data. We have applied our system to experiments on hexokinase from a variety of sources: yeast, ascites and muscle. Our system accepts qualitative stylized descriptions of experimental data, infers constraints from the observed data behaviour and then compares the experimentally inferred constraints with corresponding theoretical model-based constraints. It is desirable to have large data sets which include the results of a variety of experiments. Human intervention is needed to interpret non-kinetic information, differences in conditions, etc. Different strategies were used by the several experimenters whose data was studied to formulate mechanisms for their enzyme preparations, including different methods (product inhibitors or alternate substrates), different experimental protocols (monitoring enzyme activity differently), or different experimental conditions (temperature, pH or ionic strength). The different ordered and rapid-equilibrium mechanisms proposed by these experimenters were generally consistent with their data. On comparing the constraints derived from the several experimental data sets, they are found to be in much less disagreement than the mechanisms published, and some of the disagreement can be ascribed to different experimental conditions (especially ionic strength).

Theory formation in postulating enzyme kinetic mechanisms: reasoning with constraints.

This paper reports on a prototype system for modeling and analyzing the expert reasoning involved in postulating enzyme kinetic mechanisms. It involves data-driven, theory-driven, and analogical components of reasoning within a generate-and-test cycle. Its central component is a set of domain-specific "filters" for matching experimentally and theoretically derived constraints. The input to the system consists of an abstracted qualitative description of an experiment and prior knowledge reported in the literature. Its output shows how the results match those expected for a set of postulated reaction mechanism models and also provides a trace of which features do or do not match each of the candidate topological models. Results, constraints, and models are all analyzed and compared to those from other, similar experiments. We deduced rules for interpreting the qualitative features of enzyme kinetic experiments from natural language descriptions in the literature and verified that the rules were correct by predicting the results for typical mechanisms. We obtained the correct behavior for all 37 states of a complex enzyme mechanism involving three substrates and three products. We tested our system on data from several published reports dealing with the enzyme hexokinase and obtained detailed listings of the differences in conclusions and interpretation reported in several journal articles. This system, which provides qualitative representations of enzyme kinetic results, should facilitate further experimentation on theory formation in enzyme kinetics and lead to more efficient experimental designs.

Modeling and artificial intelligence approaches to enzyme systems.

Modeling is a means of formulating and testing complex hypotheses. Useful modeling is now possible with biological laboratory microcomputers with which experimenters feel comfortable. Artificial intelligence (AI) is sufficiently similar to modeling that AI techniques, now becoming usable on microcomputers, are applicable to modeling. Microcomputer and AI applications to physiological system studies with multienzyme models and with kinetic models of isolated enzymes are described. Using an IBM PC microcomputer, we have been able to fit kinetic enzyme models; to extend this process to design kinetic experiments by determining the optimal conditions; and to construct an enzyme (hexokinase) kinetics data base. We have also used a PC to do most of the constructing of complex multienzyme models, initially with small simple BASIC programs; alternative methods with standard spreadsheet or data base programs have been defined. Formulating and solving differential equations in appropriate representational languages, and sensitivity analysis, are soon likely to be feasible with PCs. Much of the modeling process can be stated in terms of AI expert systems, using sets of rules for fitting and evaluating models and designing further experiments. AI techniques also permit critiquing and evaluating the data, experiments, and hypotheses being modeled, and can be extended to supervise the calculations involved.