MOTIVATION: The ability to accurately read the order of nucleotides in DNA and RNA is fundamental for modern biology. Errors in next-generation sequencing can lead to many artifacts, from erroneous genome assemblies to mistaken inferences about RNA editing. Uneven coverage in datasets also contributes to false corrections. RESULT: We introduce Trowel, a massively parallelized and highly efficient error correction module for Illumina read data. Trowel both corrects erroneous base calls and boosts base qualities based on the k-mer spectrum. With high-quality k-mers and relevant base information, Trowel achieves high accuracy for different short read sequencing applications.The latency in the data path has been significantly reduced because of efficient data access and data structures. In performance evaluations, Trowel was highly competitive with other tools regardless of coverage, genome size read length and fragment size. AVAILABILITY AND IMPLEMENTATION: Trowel is written in C++ and is provided under the General Public License v3.0 (GPLv3). It is available at http://trowel-ec.sourceforge.net. CONTACT: [email protected] or [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

In this paper, we propose a rank-based image transformation scheme which is a pre-processing method for enabling more efficient compression of gray-level images by entropy encoder. For this, before we do entropy encoding on a stream of gray-level values in an image, the proposed method counts co-occurrence frequencies for neighboring pixel values. Then, it replaces each gray value with particularly ordered numbers based on the co-occurrence frequencies. Finally, the ordered number are transmitted to an entropy encoder. The pre-processing step enhances the statistical characteristic of the image transformation and thus improves the performance of entropy coding considerably. The result from our simulation using 8 bits gray-scale images shows that the proposed method can reduce bit rate by up to 37.85% compared with existing plain entropy coders.