%0 Journal Article %1 Liu:2014:Bioinformatics:24319002 %A Liu, Y %A Zhou, J %A White, K P %D 2014 %J Bioinformatics %K experimental-design replicate rna-seq sequencing-depth statistics %N 3 %P 301-304 %R 10.1093/bioinformatics/btt688 %T RNA-seq differential expression studies: more sequence or more replication? %U https://www.ncbi.nlm.nih.gov/pubmed/24319002?dopt=Abstract %V 30 %X RNA-seq is replacing microarrays as the primary tool for gene expression studies. Many RNA-seq studies have used insufficient biological replicates, resulting in low statistical power and inefficient use of sequencing resources.We show the explicit trade-off between more biological replicates and deeper sequencing in increasing power to detect differentially expressed (DE) genes. In the human cell line MCF7, adding more sequencing depth after 10 M reads gives diminishing returns on power to detect DE genes, whereas adding biological replicates improves power significantly regardless of sequencing depth. We also propose a cost-effectiveness metric for guiding the design of large-scale RNA-seq DE studies. Our analysis showed that sequencing less reads and performing more biological replication is an effective strategy to increase power and accuracy in large-scale differential expression RNA-seq studies, and provided new insights into efficient experiment design of RNA-seq studies.The code used in this paper is provided on: http://home.uchicago.edu/∼jiezhou/replication/. The expression data is deposited in the Gene Expression Omnibus under the accession ID GSE51403.

@article{Liu:2014:Bioinformatics:24319002, abstract = {RNA-seq is replacing microarrays as the primary tool for gene expression studies. Many RNA-seq studies have used insufficient biological replicates, resulting in low statistical power and inefficient use of sequencing resources.We show the explicit trade-off between more biological replicates and deeper sequencing in increasing power to detect differentially expressed (DE) genes. In the human cell line MCF7, adding more sequencing depth after 10 M reads gives diminishing returns on power to detect DE genes, whereas adding biological replicates improves power significantly regardless of sequencing depth. We also propose a cost-effectiveness metric for guiding the design of large-scale RNA-seq DE studies. Our analysis showed that sequencing less reads and performing more biological replication is an effective strategy to increase power and accuracy in large-scale differential expression RNA-seq studies, and provided new insights into efficient experiment design of RNA-seq studies.The code used in this paper is provided on: http://home.uchicago.edu/∼jiezhou/replication/. The expression data is deposited in the Gene Expression Omnibus under the accession ID GSE51403.}, added-at = {2017-10-01T09:14:49.000+0200}, author = {Liu, Y and Zhou, J and White, K P}, biburl = {https://www.bibsonomy.org/bibtex/2a242cab70ae0558586709c2c9e6185e4/marcsaric}, description = {RNA-seq differential expression studies: more sequence or more replication? - PubMed - NCBI}, doi = {10.1093/bioinformatics/btt688}, interhash = {38e8d10f3f2b419f636032b99e381cb3}, intrahash = {a242cab70ae0558586709c2c9e6185e4}, journal = {Bioinformatics}, keywords = {experimental-design replicate rna-seq sequencing-depth statistics}, month = feb, number = 3, pages = {301-304}, pmid = {24319002}, timestamp = {2017-10-01T09:14:49.000+0200}, title = {RNA-seq differential expression studies: more sequence or more replication?}, url = {https://www.ncbi.nlm.nih.gov/pubmed/24319002?dopt=Abstract}, volume = 30, year = 2014 }