The impact of the quality filter for RNA-Seq data over differential expression profile

The advent of new genome sequencing platforms in 2005, generally named nextgeneration sequencers (NGS), has boosted the number of complete genomes deposited in public databases, mainly due to the ability of those platforms to generate a huge volume of genomic data in a faster and cheaper fashion than the previous methodologies. Concomitantly, gene expression analyses have also been favored due to the development of RNA-Seq technique [1], which evaluates the whole transcriptional profile of an organism and is also used to identify new transcripts and to correct genome annotations. More interesting, the genome sequencing is not required to be finished in order to perform RNA-Seq analyses, like in Real Time technologies [2]. Data generated by high-throughput platforms are normally submitted to correction of sequence errors and read removal using quality filter in order to increase the accuracy and prevent errors during sequence assembling [3,4]. The effects of bad quality reads have already been addressed, showing the importance of removing bad quality sequences before processing the data [4]. In genome assembling, the different types of quality filters affect the proper representation of coding sequences due to the change in sequencing coverage [5]. Considering that the sequencing depth is used to measure the gene expression on cDNA sequencing (RNA-seq), the removal of reads through the use of quality filters may affect the gene transcriptional profile. In this work, we evaluate the effects of applying different quality filters (Phred) in gene expression analyses of Corynebacterium pseudotuberculosis 1002. The rRNA was extracted from Corynebacterium pseudotubeculosis 1002 under four conditions: control, heat shock (50°C), osmotic stress (2M NaCl) and acidic stress (pH). The cDNA was then synthesized, and sequenced in SOLID platform. The generated data was submitted to three different quality filters (QV10, QV15 and QV20) and also analyzed in a standard condition, without quality filter (QV0). After