Comment on “circ_0000317/microrna‐520g/hoxd10 Axis Affects the Biological Characteristics of Colorectal Cancer”

Yuqing Wang,Xingqiang Gao,Liu Wei,Lei Yan

DOI: https://doi.org/10.1002/kjm2.12466

2021-01-01

Abstract:Noncoding RNAs (ncRNAs) act as key regulators of physiological programs in developmental and disease contexts, which have been identified as oncogenic drivers and tumor suppressors in some types of cancer.1 We read with great interest the article by Lai FJ and colleagues (Lai FJ, et al. Kaohsiung J Med Sci. 2021) who found that the circular RNA circ_0000317 repressed colorectal cancer progression by targeting miR-520g and modulating HOXD10 expression.2 This finding has far-reaching implications. However, we believe it is necessary to further clarify the bioinformatics analysis strategy used in this study. First, from the description in their paper, it seems that the raw microarray data were not preprocessed. However, we are of the opinion that these steps are necessary because data preprocessing is an important guarantee of the reliability of microarray analysis results. Second, according to the authors' description, it seems that unadjusted p values and gene expression fold change values were used to define significantly differentially expressed circular RNAs. However, because of the likelihood of high false positives caused by multiple comparisons, the statistical methods used for microarray analysis should be carefully selected. Only accurate analysis results will provide a convincing basis for subsequent experiments. In order to deal with the problem of multiple comparisons, we suggest that Benjamini–Hochberg or Bonferroni adjusted p values can be used. In addition, linear modeling with empirical Bayes moderation showed good control of the false discovery rate and reasonable sensitivity when defining differentially expressed noncoding RNAs.3 We consider that the R/Bioconductor software package Linear Models for Microarray Analysis (Limma), which uses linear models to analyze microarray and high-throughput PCR data, is the preferred method for such analyses.4, 5 Limma results can be used to choose appropriate expression fold changes and false discovery rates <0.05 as conservative cutoffs to analyze the changes in gene expression. The authors acknowledge Margaret Biswas, PhD, for editing the English text of a draft of this manuscript. All authors declare no conflict of interest.