Prediction of Growth Parameters of In Vitro Microtubules in Association with Different Isoforms of Tau

Microtubules are cellular organelles that are important for many cellular processes including cell shape and cell division. Microtubules grow and shorten by a process called the dynamic instability. The dynamics of microtubules can be regulated by different microtubule associated proteins such as tau. Predicting the behavior of microtubules for a hitherto unseen condition of tau from a database of already performed experiments with other conditions of tau will guide the biologists in performing future experiments and exploring unusual values. We employ two different methods of prediction – nearest neighbor and surface fitting. The nearest neighbor method first looks at the neighbors of a condition and its relationship with them. It then applies the average change of all similar relationships in the database to the neighbors to predict. The surface fitting method finds the equation of a surface that best describes the database values and then uses that surface to predict the value for the unseen condition. Experiments were performed with in vitro data with different variants of tau. Since growth is the most important phenomenon for this type of data, two growth parameters were predicted: i) the median growth rate of the microtubules and ii) the percentage of time the microtubules grow. The relative differences between the predicted and the actual values were at most 2%.