Given the overwhelming impact of machine learning on the last decade, several libraries and frameworks have been developed in recent years to simplify the design and training of neural networks, providing array-based programming, automatic differentiation and user-friendly access to hardware accelerators. None of those tools, however, was designed with native and transparent support for Cloud Computing or heterogeneous High-Performance Computing (HPC). The DeepHealth Toolkit is an open source Deep Learning toolkit aimed at boosting productivity of data scientists operating in the medical field by providing a unified framework for the distributed training of neural networks, which is able to leverage hybrid HPC and cloud environments in a transparent way for the user. The toolkit is composed of a Computer Vision library, a Deep Learning library, and a front-end for non-expert users; all of the components are focused on the medical domain, but they are general purpose and can be applied to any other field. In this paper, the principles driving the design of the DeepHealth libraries are described, along with details about the implementation and the interaction between the different elements composing the toolkit. Finally, experiments on common benchmarks prove the efficiency of each separate component and of the DeepHealth Toolkit overall.

SUMMARY: BioBlend.objects is a new component of the BioBlend package, adding an object-oriented interface for the Galaxy REST-based application programming interface. It improves support for metacomputing on Galaxy entities by providing higher-level functionality and allowing users to more easily create programs to explore, query and create Galaxy datasets and workflows. AVAILABILITY AND IMPLEMENTATION: BioBlend.objects is available online at https://github.com/afgane/bioblend. The new object-oriented API is implemented by the galaxy/objects subpackage.

SUMMARY: SEAL is a scalable tool for short read pair mapping and duplicate removal. It computes mappings that are consistent with those produced by BWA and removes duplicates according to the same criteria employed by Picard MarkDuplicates. On a 16-node Hadoop cluster, it is capable of processing about 13 GB per hour in map+rmdup mode, while reaching a throughput of 19 GB per hour in mapping-only mode. AVAILABILITY: SEAL is available online at http://biodoop-seal.sourceforge.net/.

MapReduce has become increasingly popular as a simple and efficient paradigm for large-scale data processing. One of the main reasons for its popularity is the availability of a production-level open source implementation, Hadoop, written in Java. There is considerable interest, however, in tools that enable Python programmers to access the framework, due to the language's high popularity. Here we present a Python package that provides an API for both the MapReduce and the distributed file system sections of Hadoop, and show its advantages with respect to the other available solutions for Hadoop Python programming, Jython and Hadoop Streaming.

The rapid growth in the throughput to cost ratio of experimental data production technologies is generating vast amounts of scientific data, often organized into "large" objects (genomes, bio-images) exhibiting complex internal structures. Frequently, datasets must be shared between multiple research groups interested not only in the final results, but also in how they are produced. The practical difficulties of moving terabytes or more of data across the network, as well as the need to maintain a clear separation between software stack and storage infrastructure, are thus raising interest in the use of virtual clusters for HPC and data intensive applications. In this paper we employ a MapReduce implementation of an image analysis pipeline used by deep sequencing platforms to analyse different virtual cluster scenarios and their impact on system performance.