Tool recommendation for workflow composition using frequent patterns

Workflows or pipelines provide a means for executing complex data analysis seamlessly. Composing tools into a workflow is essential in bioinformatics experiments. There are scientific workflow systems such as Taverna and Galaxy that facilitate automatic workflow composition. However, designing workflows using workflow systems becomes more complex with the availability of vast numbers of complex, heterogeneous tools. Connecting such heterogeneous tools to a workflow is error-prone and time-consuming. The objective of the study is to develop a suggestive system for interactive workflow composition using frequent patterns in workflows. The approach basically consists of three main phases: pattern mining, component suggestion, and updating the workflow. Frequent patterns of workflows are identified using frequent subgraph mining techniques and N-gram modeling. The suggested components allow reusing best practice workflows while reducing the time required in composing the workflows. Frequent usage patterns identified can also be used in searching similar workflows in workflow repositories. An interactive workflow composition approach is useful for novice as well as experienced scientists in composing workflows with state-of-the-art tools. The approach enhances the reuse of best practice workflows developed by other users. Such systems would succeed more in future with the availability of more and more workflows in the light of open science