Dataset Information

Trimming of sequence reads alters RNA-Seq gene expression estimates

ABSTRACT: Background High-throughput RNA-Sequencing (RNA-Seq) has become the preferred technique for studying gene expression differences between biological samples and for discovering novel isoforms, though the techniques to analyze the resulting data are still immature. One pre-processing step that is widely but heterogeneously applied is trimming, in which low quality bases, identified by the probability that they are called incorrectly, are removed. However, the impact of trimming on subsequent alignment to a genome could influence downstream analyses including gene expression estimation; we hypothesized that this might occur in an inconsistent manner across different genes, resulting in differential bias. Results To assess the effects of trimming on gene expression, we generated RNA-Seq data sets from four samples of larval Drosophila melanogaster sensory neurons, and used three trimming algorithms—SolexaQA, Trimmomatic, and ConDeTri—to perform quality-based trimming across a wide range of stringencies. After aligning the reads to the D. melanogaster genome with TopHat2, we used Cuffdiff2 to compare the original, untrimmed gene expression estimates to those following trimming. With the most aggressive trimming parameters, over ten percent of genes had significant changes in their estimated expression levels. This trend was seen with two additional RNA-Seq data sets and with alternative differential expression analysis pipelines. We found that the majority of the expression changes could be mitigated by imposing a minimum length filter following trimming, suggesting that the differential gene expression was primarily being driven by spurious mapping of short reads. Slight differences with the untrimmed data set remained after length filtering, which were associated with genes with low exon numbers and high GC content. Finally, an analysis of paired RNA-seq/microarray data sets suggests that no or modest trimming results in the most biologically accurate gene expression estimates. Conclusions We find that aggressive quality-based trimming has a large impact on the apparent makeup of RNA-Seq-based gene expression estimates, and that short reads can have a particularly strong impact. We conclude that implementation of trimming in RNA-Seq analysis workflows warrants caution, and if used, should be used in conjunction with a minimum read length filter to minimize the introduction of unpredictable changes in expression estimates.

ORGANISM(S): Drosophila melanogaster

PROVIDER: GSE72884 | GEO | 2016/02/25

SECONDARY ACCESSION(S): PRJNA295261

REPOSITORIES: GEO

ACCESS DATA

Dataset's files

Source:

			Action	DRS
		Other

Items per page:

1 - 1 of 1

Similar Datasets

Project description:Background High-throughput RNA-Sequencing (RNA-Seq) has become the preferred technique for studying gene expression differences between biological samples and for discovering novel isoforms, though the techniques to analyze the resulting data are still immature. One pre-processing step that is widely but heterogeneously applied is trimming, in which low quality bases, identified by the probability that they are called incorrectly, are removed. However, the impact of trimming on subsequent alignment to a genome could influence downstream analyses including gene expression estimation; we hypothesized that this might occur in an inconsistent manner across different genes, resulting in differential bias. Results To assess the effects of trimming on gene expression, we generated RNA-Seq data sets from four samples of larval Drosophila melanogaster sensory neurons, and used three trimming algorithmsâSolexaQA, Trimmomatic, and ConDeTriâto perform quality-based trimming across a wide range of stringencies. After aligning the reads to the D. melanogaster genome with TopHat2, we used Cuffdiff2 to compare the original, untrimmed gene expression estimates to those following trimming. With the most aggressive trimming parameters, over ten percent of genes had significant changes in their estimated expression levels. This trend was seen with two additional RNA-Seq data sets and with alternative differential expression analysis pipelines. We found that the majority of the expression changes could be mitigated by imposing a minimum length filter following trimming, suggesting that the differential gene expression was primarily being driven by spurious mapping of short reads. Slight differences with the untrimmed data set remained after length filtering, which were associated with genes with low exon numbers and high GC content. Finally, an analysis of paired RNA-seq/microarray data sets suggests that no or modest trimming results in the most biologically accurate gene expression estimates. Conclusions We find that aggressive quality-based trimming has a large impact on the apparent makeup of RNA-Seq-based gene expression estimates, and that short reads can have a particularly strong impact. We conclude that implementation of trimming in RNA-Seq analysis workflows warrants caution, and if used, should be used in conjunction with a minimum read length filter to minimize the introduction of unpredictable changes in expression estimates. Four biological replicates of 100 Drosophila melanogaster larval multi-dendritic sensory neurons were profiled by mRNA-Seq

Project description:Small RNA sequencing has been performed using eight commonly used breast cancer cell lines by Illumina as part of the Illumina iDEA Challange 2011. Total RNA of the cell lines was isolated and the sequencing library was prepared using Illuminas Small RNA v1.5 library prep protocol. This data was analyzed with Flicker, an add-on to the Illumina Genome Analyzer Pipeline software that allows for the preliminary analysis of small RNA runs on the GA. Flicker was used in this case for: * Trimming: Attempts to trim the adaptor sequence from small RNA reads Trimming is accomplished by aligning the adaptor with each read, both internally (the adaptor is fully contained within the read) and as a suffix (the adaptor overlaps the 3' end of the read). A similarity matrix is used to generate the matching score, taking into account both substitutions and indels. The alignment with the best matching score is returned, and that best alignment is trimmed from the read. A trimming threshold may be specified, and the adaptor is not trimmed unless the best score exceeds the threshold. * Aligning with Iterated ELAND: Attempts to aligns trimmed reads to targets (squashed genomes or databases such as miRBase) using the ELAND short read aligner. Each read is aligned by ELAND with specified alignment targets. ELAND was designed to align reads of a uniform length, and the trimmed reads from a small RNA run are several different lengths. We currently address this by running ELAND for reads of each length: reads are binned by length, and each bin of a given length is aligned against a specified target. Once all bins are aligned against all targets (in this case, gDNA and miRBase), the results are recombined into a single file with all the annotation information. ===== Key to Experiments HCT ID, Lane, Sample ID, Cell Line, ER +/- Status, Reads HCT20061, Lane 1, HS378, MCF7, +, 14,555,390 HCT20062, Lane 2, HS379, MDA-MB-231, -, 15,147,037 HCT20063, Lane 3, HS381, T47D, +, 14,701,254 HCT20064, Lane 4, HS377, BT-20, -, 16,529,823 HCT20065, Lane 5, HS375, BT-474, +, 16,443,361 HCT20066, Lane 6, HS380, MDA-MB-468, -, 16,746,981 HCT20067, Lane 7, HS382, ZR-75-1, +, 17,631,005 HCT20068, Lane 8, HS376, MCF10A, normal, 19,155,607

Dataset Information

Trimming of sequence reads alters RNA-Seq gene expression estimates

Dataset's files

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets