MiAMi: Microbiome Analysis using signatures for undocumented Microbes


Principal investigator: Prof. Giri Narasimhan

Pipeline architect: Dr. Trevor Cickovski



One of the important challenges in analyzing microbiome data is to create flexible lightweight pipelines using a variety of heterogeneous software tools for pipeline stages. These tools can be constructed by third parties using their choice of both language and user interaction, complicating integration of these tools. Here we use the concept of Plugins to build a lightweight pipeline that can accommodate and integrate these heterogeneous tools, independent of these choices. We validate the framework by integrating tools applied at different stages of the microbiome analysis pipeline, implemented in both scripting and compiled languages for both the CPU and GPU.



Cickovski, T., Aguiar-Pulido, V., Huang, W., Mahmud, S., Narasimhan, G. (2016). Lightweight Microbiome Analysis Pipelines, Accepted at IWBBIO


Our lightweight computational backend is implemented in C++ and dynamically loads and executes user-specified plugins step-by-step. A configuration file (called "config.txt") contains all plugins for the pipeline. Plugins may be written in different languages, including Python, CUDA, C++, R, Perl. We have provided more than one example configuration file, each implementing different pipelines using different sets of plugins. We have also provided a README file that explains how to compile and run our sample pipelines with the help of SCons (can be downloaded from Scons) Click below to download the tar ball.