Dataset Information

Calculating sample size estimates for RNA sequencing data.

ABSTRACT: Given the high technical reproducibility and orders of magnitude greater resolution than microarrays, next-generation sequencing of mRNA (RNA-Seq) is quickly becoming the de facto standard for measuring levels of gene expression in biological experiments. Two important questions must be taken into consideration when designing a particular experiment, namely, 1) how deep does one need to sequence? and, 2) how many biological replicates are necessary to observe a significant change in expression?Based on the gene expression distributions from 127 RNA-Seq experiments, we find evidence that 91% ± 4% of all annotated genes are sequenced at a frequency of 0.1 times per million bases mapped, regardless of sample source. Based on this observation, and combining this information with other parameters such as biological variation and technical variation that we empirically estimate from our large datasets, we developed a model to estimate the statistical power needed to identify differentially expressed genes from RNA-Seq experiments.Our results provide a needed reference for ensuring RNA-Seq gene expression studies are conducted with the optimally sample size, power, and sequencing depth. We also make available both R code and an Excel worksheet for investigators to calculate for their own experiments.

SUBMITTER: Hart SN

PROVIDER: S-EPMC3842884 | biostudies-other | 2013 Dec

REPOSITORIES: biostudies-other

ACCESS DATA

Publications

Calculating sample size estimates for RNA sequencing data.

Hart Steven N SN Therneau Terry M TM Zhang Yuji Y Poland Gregory A GA Kocher Jean-Pierre JP

Journal of computational biology : a journal of computational molecular cell biology 20130820 12

<h4>Background</h4>Given the high technical reproducibility and orders of magnitude greater resolution than microarrays, next-generation sequencing of mRNA (RNA-Seq) is quickly becoming the de facto standard for measuring levels of gene expression in biological experiments. Two important questions must be taken into consideration when designing a particular experiment, namely, 1) how deep does one need to sequence? and, 2) how many biological replicates are necessary to observe a significant cha ...[more]

PMID: 23961961

Similar Datasets

Project description:Sample size is a critical aspect of study design in population genomics research, yet few empirical studies have examined the impacts of small sample sizes. We used datasets from eight diverging bird lineages to make pairwise comparisons at different levels of taxonomic divergence (populations, subspecies, and species). Our data are from loci linked to ultraconserved elements and our analyses used one single nucleotide polymorphism per locus. All individuals were genotyped at all loci, effectively doubling sample size for coalescent analyses. We estimated population demographic parameters (effective population size, migration rate, and time since divergence) in a coalescent framework using Diffusion Approximation for Demographic Inference, an allele frequency spectrum method. Using divergence-with-gene-flow models optimized with full datasets, we subsampled at sequentially smaller sample sizes from full datasets of 6-8 diploid individuals per population (with both alleles called) down to 1:1, and then we compared estimates and their changes in accuracy. Accuracy was strongly affected by sample size, with considerable differences among estimated parameters and among lineages. Effective population size parameters (ν) tended to be underestimated at low sample sizes (fewer than three diploid individuals per population, or 6:6 haplotypes in coalescent terms). Migration (m) was fairly consistently estimated until <2 individuals per population, and no consistent trend of over-or underestimation was found in either time since divergence (T) or theta (Θ = 4N refμ). Lineages that were taxonomically recognized above the population level (subspecies and species pairs; that is, deeper divergences) tended to have lower variation in scaled root mean square error of parameter estimation at smaller sample sizes than population-level divergences, and many parameters were estimated accurately down to three diploid individuals per population. Shallower divergence levels (i.e., populations) often required at least five individuals per population for reliable demographic inferences using this approach. Although divergence levels might be unknown at the outset of study design, our results provide a framework for planning appropriate sampling and for interpreting results if smaller sample sizes must be used.

Project description:ObjectiveTo assess the influence of trial sample size on treatment effect estimates within meta-analyses.DesignMeta-epidemiological study.Data sources93 meta-analyses (735 randomised controlled trials) assessing therapeutic interventions with binary outcomes, published in the 10 leading journals of each medical subject category of the Journal Citation Reports or in the Cochrane Database of Systematic Reviews.Data extractionSample size, outcome data, and risk of bias extracted from each trial.Data synthesisTrials within each meta-analysis were sorted by their sample size: using quarters within each meta-analysis (from quarter 1 with 25% of the smallest trials, to quarter 4 with 25% of the largest trials), and using size groups across meta-analyses (ranging from <50 to ≥ 1000 patients). Treatment effects were compared within each meta-analysis between quarters or between size groups by average ratios of odds ratios (where a ratio of odds ratios less than 1 indicates larger effects in smaller trials).ResultsTreatment effect estimates were significantly larger in smaller trials, regardless of sample size. Compared with quarter 4 (which included the largest trials), treatment effects were, on average, 32% larger in trials in quarter 1 (which included the smallest trials; ratio of odds ratios 0.68, 95% confidence interval 0.57 to 0.82), 17% larger in trials in quarter 2 (0.83, 0.75 to 0.91), and 12% larger in trials in quarter 3 (0.88, 0.82 to 0.95). Similar results were obtained when comparing treatment effect estimates between different size groups. Compared with trials of 1000 patients or more, treatment effects were, on average, 48% larger in trials with fewer than 50 patients (0.52, 0.41 to 0.66) and 10% larger in trials with 500-999 patients (0.90, 0.82 to 1.00).ConclusionsTreatment effect estimates differed within meta-analyses solely based on trial sample size, with stronger effect estimates seen in small to moderately sized trials than in the largest trials.

Dataset Information

Calculating sample size estimates for RNA sequencing data.

Publications

Calculating sample size estimates for RNA sequencing data.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets