We read with great interest the recent report by Semeghini et al. (2017), Cell Biol Int, "Menopause transition promotes distinct modulation of mRNAs and miRNAs expression in calvaria and bone marrow osteoblastic cells," which appeared on 24 May 2017 in Cell Biology International. The results of the report are very helpful for us; however, from our perspective, the author's method in bioinformatics analysis is inappropriate: Student's t-test is an inappropriate statistical method for detecting differentially expressed mRNA or miRNA in osteoblastic cells from calvaria of ovariectomized rats compared to control or in osteoblastic cells from bone marrow of ovariectomized rats compared to control.Keywords: computational methods; theoretical biology Dear editor:We read with great interest the recent report by Semeghini et al. (2017), "Menopause transition promotes distinct modulation of mRNAs and miRNAs expression in calvaria and bone marrow osteoblastic cells," which appeared on 24 May 2017 in Cell Biology International. The results of the report are very helpful for us; however, from our perspective, the author's method in bioinformatics analysis is inappropriate.We noticed that the authors used Student's t-tests for detecting differentially expressed genes (DEGs) in osteoblastic cells from calvaria of ovariectomized rats compared or from bone marrow of ovariectomized rats compared to control. Affectively, due to the high false positive caused by a huge number of probes and multiple comparisons, it is fundamental to analyze microarray data properly to reach a reliable result by rational statistical method, but Student's t-test is not suitable for high-level microarray analysis. If Student's t-test was used for differentially expressed genes analysis in microarray, setting P-value <0.01 as cutoff is the correct method. We recommend using limma (Linear Models for Microarray Analysis) (Law et al., 2016), a commonly used statistical test to analyze differential expression package by using linear models, and choosing more than 1.5-fold expression changes and false discovery rate (FDR) <0.05 as cutoff is an appropriate and conservative approach to obtain DEGs. Moreover, Significant Analysis of Microarray (SAM) (Tusher et al., 2001) is also a considerable non-parametric statistical algorithm, and twofold expression change and q < 0.1 is rational cutoff to obtain DEGs.Above all, although the authors preformed extra analysis for a portion of DEGs in order to verify the results of the microarray, it is impractical using PCR or other technology to verify all DEGs. Choosing the right statistical method (Chrominski and Tkacz, 2015) and obtaining more accurate and convincing results of DEGs analysis is the basis for further analysis such as GO enrichment analysis and KEGG pathway analysis.
ReferencesChrominski K, Tkacz M (2015) Comparison of high-level microarray analysis methods in the context of result consistency [J]. PLoS ONE 10(6): 0128845. Law CW, Alhamdoosh M, Su S, Smyth GK, Ritchie ME (2016) RNA-seq analysis is e...