Marchaukoski, J. N. | Search bioRxiv

bioRxiv2018

RAFTS3G - An efficient and versatile clustering software to analyses in large protein datasets

The need to develop computational tools and techniques that can predict efficiently consistent groups of family proteins in large volume of biological information is still a great perspective in Bioinformatic studies. Besides that, it is difficult to increase speed demanding low computational processing to minimize the information complexity. Tools already consolidated as the CD-HIT and UCLUST generates very compact data that makes the Data Mining difficult and have low efficiency when used for detect homology among proteins requiring manual intervention, therefore it is necessary a tool that is also efficient in low similarity. Here we present a new approach for the Data Mining and analysis of homology in large dataset of protein sequences, the RAFTS3G. We used the UniProtKB/Swiss-Prot database with the most popular clustering tools and RAFTS3G proved to be more than 10 times faster than CD-HIT and its strategy increases the performance in low similarity to detect protein families.\n\nContact: raittz@ufpr.br

bioinformatics↗

bioRxiv2018

The identification of DNA binding regions of the σ54 factor using artificial neural network

Transcription of many bacterial genes is regulated by alternative RNA polymerase sigma factors as the sigma 54 ({sigma}54). A single essential {sigma} promotes transcription of thousands of genes and many alternative {sigma} factors promote transcription of multiple specialized genes required for coping with stress or development. Bacterial genomes have two families of sigma factors, sigma 70 ({sigma}70) and sigma 54 ({sigma}54). {sigma}54 uses a more complex mechanism with specialized enhancers-binding proteins and DNA melting and is well known for its role in regulation of nitrogen metabolism in proteobacteria. The identification of these regulatory elements is the main step to understand the metabolic networks. In this study, we propose a supervised pattern recognition model with neural network to identify Transcription Factor Binding Sites (TFBSs) for {sigma}54. This approach is capable of detecting {sigma}54 TFBSs with sensitivity higher than 98% in recent published data. False positives are reduced with the addition of ANN and feature extraction, which increase the specificity of the program. We also propose a free, fast and friendly tool for {sigma}54 recognition and a {sigma}54 related genes database, available for consult. S54Finder can analyze from short DNA sequences to complete genomes and is available online. The software was used to determine {sigma}54 TFBSs on the complete bacterial genomes database from NCBI and the result is available for comparison. S54Finder does the identification of {sigma}54 regulated genes for a large set of genomes allowing evolutionary and conservation studies of the regulation system between the organisms.

bioinformatics↗