What is the architecture of BiotextMiner?
The architecture of BiotextMiner is made highly modular, flexible and scalable keeping end users in mind. For example, for analysis a user has the flexibility to choose our training/test data, or he may choose to have these of his own. In addition, if a user wishes to use his own protein name and interaction keyword dictionaries (instead of those provided as default) he can get in touch with us for a quick customized solution. The overall workflow schematic of the webserver is shown in Figure 1 below.
The architecture of BiotextMiner is made highly modular, flexible and scalable keeping end users in mind. For example, for analysis a user has the flexibility to choose our training/test data, or he may choose to have these of his own. In addition, if a user wishes to use his own protein name and interaction keyword dictionaries (instead of those provided as default) he can get in touch with us for a quick customized solution. The overall workflow schematic of the webserver is shown in Figure 1 below. The webserver has two main independent modules: • Module A (testing of labeled training and test datasets): this module can be used by a user to test the accuracy of our PI system/model using his or our (default) training or test datasets or both. The output/result from Module A has three components, namely i) Training data accuracy based on10-fold cross validation, ii) Users test data accuracy, and iii) Individual predictions on labeled samples in the users test data. If any (or both) of
