Adjustment of RNASeq data for the effect of highly abundant transcripts: a case study in milk production (622.4)
Academic Article
Overview
Research
Identity
Additional Document Info
Other
View All
Overview
abstract
During lactation, profound changes occur in the mammary gland and thousands of genes undergo differential regulation. RNAseq data analysis reveals a tiny minority of milk genes account for the vast majority of total gene expression. This striking imbalance in transcript abundances poses a significant problem for data analysis, e.g. low abundance genes may be incorrectly identified as downregulated. To tackle this problem, we developed a Dilution Adjustment Model which more accurately classifies changes in levels of low abundance transcripts between transcriptomes at two developmental stages (baseline and mature lactation). Applying this model to human and bovine milk data led us to reclassify 2,155 human (971 bovine) genes as upregulated instead of not differentially expressed and 2,524 human (1,732 bovine) as unregulated instead of downregulated. Changes in gene classification were supported by analysis of Gene Ontology and TFBS enrichment profiles. Investigation of ChIPSeq data showed genes reclassified as upregulated exhibit signs of active transcription and reclassified unregulated genes match marks of others in this set. This work leads to a better understanding of transcription mechanisms in milk production. Application of this model to other biological systems is relevant where transcripts from a minority of genes dominate transcriptome composition.Grant Funding Source: K.B was supported by Grant Number T32GM008799 from NIGMSNIH
