Despite the potential wealth of educational indicators expressed in a student's approach to homework assignments, how students arrive at their final solution is largely overlooked in university courses. In this paper we present a methodology which uses machine learning techniques to autonomously create a graphical model of how students in an introductory programming course progress through a homework assignment. We subsequently show that this model is predictive of which students will struggle with material presented later in the class.

New possibilities in online education create new challenges.

A homology derived molecular model of prostate specific antigen (PSA) was created and refined. The active site region was investigated for specific interacting functionality and a binding model postulated for the novel 2-azetidinone acyl enzyme inhibitor 1 (IC(50) = 8.98 +/- 0.90 microM) which was used as a lead compound in this study. A single low energy conformation structure II (Figure 2) was adopted as most likely to represent binding after minimization and dynamics calculations. Systematic analysis of the binding importance of all three side chains appended to the 2-azetidinone was conducted by the synthesis of several analogues. A proposed salt bridge to Lys-145 with 4 (IC(50) = 5.84 +/- 0.92 microM) gave improved inhibition, but generally the binding of the N-1 side chain in a specific secondary aromatic binding site did not tolerate much structural alteration. A hydrophobic interaction of the C-4 side chain afforded inhibitor 6 (IC(50) = 1.43 +/- 0.19 microM), and polar functionality could also be added in a proposed interaction with Gln-166 in 5 (IC(50) = 1.34 +/- 0.05 microM). Reversal of the C-4 ester connectivity furnished inhibitors 7 (IC(50) = 1.59 +/- 0.15 microM), 11 (IC(50) = 3.08 +/- 0.41 microM), and 13 (IC(50) = 2.19 +/- 0.36 microM) which were perceived to bind to PSA by a rotation of 180 degrees relative to the C-4 ester of normal connectivity. Incorporation of hydroxyl functionality into the C-3 side chain provided 16 (IC(50) = 348 +/- 50 nM) with the greatest increase in PSA inhibition by a single modification. Multiple copy simultaneous search (MCSS) analysis of the PSA active site further supported our model and suggested that 18 would bind strongly. Asymmetric synthesis yielded 18 (IC(50) = 226 +/- 10 nM) as the most potent inhibitor of PSA reported to date. It is concluded that our design approach has been successful in developing PSA inhibitors and could also be applied to the inhibition of other enzymes, especially in the absence of crystallographic information.

Do we need another editorial about engaging students in learning computer science so they will stay in the field and prepare for a career or further study? We wish it were not so, but in spite of some progress, there is little evidence that our students complete courses or stay in their degree programs at better rates than a few years ago. There are bright spots in the picture and some promising results here and there, but an overall pattern of change and improvement is not yet evident.