PRONTO-TK: a user-friendly PROtein Neural neTwOrk tool-kit for accessible protein function prediction

Abstract

Associating one or more Gene Ontology (GO) terms to a protein means making a statement about a particular functional characteristic of the protein. This association provides scientists with a snapshot of the biological context of the protein activity. This paper introduces PRONTO-TK, a Python-based software toolkit designed to democratize access to Neural-Network based complex protein function prediction workflows. PRONTO-TK is a user-friendly graphical interface (GUI) for empowering researchers, even those with minimal programming experience, to leverage state-of-the-art Deep Learning architectures for protein function annotation using GO terms. We demonstrate PRONTO-TK's effectiveness on a running example, by showing how its intuitive configuration allows it to easily generate complex analyses while avoiding the complexities of building such a pipeline from scratch.

Figure 1.

PRONTO-TK’s Graphical User Interface. (A) Main Control Panel: each pipeline can be enabled in full (by clicking on the run button), or step-by-step (by clicking on each individual phase). Each phase is enabled if the correct input files are present. Available files are colored in orange, whereas missing ones are colored in purple. If only some of the required files are present, the file box is colored in light gray. (B) Configuration Panel for the experiments’ configuration files. The configuration panel allows setting all the configuration parameters. A short help precedes every configuration item. (C) Log window for each step of the pipeline. A progress bar and a text box allow to track the task progress.

Figure 2.

PRONTO-TK’s visual outputs. (A). A plot of loss and accuracy versus epochs for each trained or fine-tuned model. (B-1). 3D plot summarizing the performances of the trained models. Each point corresponds to a combination of the input parameters (in this case we have 27 points corresponding to the 27 models). The color of the dot derives from the F-score of the model. Red dots have an F-score < 0.9. Green dots are shaded accordingly to their F-score (darker green is a lower F-score). The yellow dot corresponds to the model with the best F-score. Moving the mouse over a dot shows its parameters. (B-2) 3D plot of Accuracy, Precision, and Recall for all trained models. Larger points correspond to the average values (of Accuracy, Precision, and Recall) for all models trained with the same parameters for each leave-one-species-out species. Moving the mouse over a point will display the corresponding model parameters. (C). For each combination of model parameters, the violin plots of the distribution of the predicted probabilities for each label in the validation of each ‘leave-one-species-out’ species. This is repeated for true positives (TP), false positives (FP), true negatives (TN) and false negatives (FN), followed by a table reporting the measures of the predictive performances achieved by the model for each species.