Probabilistic approaches activity prediction

Before presenting our results on structural implications of the observed and predicted levels of activity in the MB and of the requirement of lossless information transmission, some explanation of our probabilistic approach seems warranted. The probability Ppn of activity in a given PN mainly reflects properties of the input space (odor space) and different patterns of PN activity are diced out for every LFP cycle. The connection probabilities pn->kc and kc bkc, on the other hand, refer to the random connectivity of each locust, i.e., the connectivity is determined only once for each animal. In building distributions (and taking averages) with respect to both probability spaces, we are making statements about the distribution of (and the typical value of) properties for all locusts in response to all possible odors, in a sense. 

Filimonov D, Poroikov V (2008) Probabilistic approach in activity prediction. In Vamek A, Tropsha A (eds) Chemoinformatics approaches virtual screen. RSC Publishing, Cambridge,... 

Simplistic and heuristic similarity-based approaches can hardly produce as good predictive models as modern statistical and machine learning methods that are able to assess quantitatively biological or physicochemical properties. QSAR-based virtual screening consists of direct assessment of activity values (numerical or binary) of all compounds in the database followed by selection of hits possessing desirable activity. Mathematical methods used for models preparation can be subdivided into classification and regression approaches. The former decide whether a given compound is active, whereas the latter numerically evaluate the activity values. Classification approaches that assess probability of decisions are called probabilistic. 

Both new pharmacotherapeutics and adverse/toxic effect can be discovered on the basis of computer predictions with probabilistic methods. Different methods can be applied either sequentially or simultaneously. Early attempts to predict many kinds of biological activity simultaneously using such an approach were performed by Avidon and co-authors,Golender and Rozenblit, " and Vassiliev and co-authors. 

---