the era of biological omics, an essential issue is how to mine the important biological information, including important signaling molecule patterns, signaling pathways, molecular networks, and pathway-network systems, from big omic data in combination with phenotype features in a given biological system. An appropriate statistical method and algorithm play central technique-support roles in such a bioinformation-mining process. However, for the statistical results, one must realize the difference and relationship of statistical vs. biological significances in those analysis processes. Statistical significance and biological significance are two different concepts with overlapping of their results. The choice of statistical method and threshold value of statistical significance should be determined by the data type and research goal. A statistically significant result must be reasonably interpreted with corresponding biological processes to decide its biological significance. One must not use statistical significance to kidnap biological significance, and the statistical result is only a reference to determine a biological significance.

The rapid development of biological omics (genomics, transcriptomics, proteomics, peptidomics, and metabolomics) [

The development of modern molecular medicine is experiencing at least three paradigm shifts (Figure 1): (i) from macrocosmic view to microcosmic view, which is from anatomy, histology, cytobiolgy, molecular biology, to structural biology. (ii) From traditional single parameter strategy to multi-parameter systematic strategy [

The paradigm shifts (ii) and (iii) mainly benefit from the rapid developments of omics (genomics, transcriptomics, proteomics, peptidomics, metabolomics, and interactomics) and systems biology together with identification of phenotype including different clinical characteristics (Figure 3) [2-5,12,13]. Furthermore, the genetic central rule (Figure 4) [

The large-scale and complicated omics data in combination with different clinical characteristics (Figures 3 and 4) must be analyzed with appropriate statistical methods to reveal important signal molecule pattern [

One example was taken below to clarify the statistical consideration and biological significance in the systems biology study. One study [

In this study, one can clearly find that if the significance level of 0.01 or 0.001 is used, it is more stringent criteria for this type of data analysis. It could decrease the probability of false positives, but it also leads to the loss of some biologically meaningful information. For example, if the significance level of 0.001 [or -log (0.001) = 3] is utilized, then there will be only 1 left significant canonical pathway (Figure 5) identified from pituitary adenoma nitroproteomic data, 10 significant canonical pathways (Figure 6) from control pituitary nitroproteomic data, no significant canonical pathways (Figure 7) derived from comparative proteomics data, and 7 statistically significant canonical pathways (Figure 8) from the protein-mapping data. As a result, compared to the level of statistical significance p < 0.05, many important canonical pathways (Figures 5-8) are missing at the level of statistical significance p < 0.01 or 0.001. In fact, many differentially expressed proteins (DEPs) with a biological significance [

Any statistically significant result only serves as a reference for biological significance, and must be rationally interpreted with corresponding biological processes to decide its biological significance. Statistical significance does not mean a real variation or effect in a biological system. Some statistically significant results do not have any real biological meaning at al. A typical example is that hemoglobin is often identified as statistically significant DEP between pituitary adenoma and control tissues [

Based on these statistical considerations, those statistically significant pathways and networks identified with the Fisher's exact test with a significance level of 0.05 were reasonably explained within the pituitary adenoma biological system. Four important biological pathways [

Molecules from genome, transcriptome, proteome, metabolome interact mutually to form an interactome to exert their biological functions in a biological system [

It is very complicated and important for statistical consideration and biological significance in a given big data and biological system. One must realize the difference and relationship between statistical and biological significances. The right statistical method must be selected for a given big data. The statistical results must be reasonably explained in a specific biological system. One must not use statistical significance to kidnap biological significance. Statistical significance is not equal to biological significance. Statistical result is only a reference to determine a biological significance.

This work was supported by the National Natural Science Foundation of China (Grant No. 81572278 and 81272798 to XZ), the grants from China “863” Plan Project (Grant No. 2014AA020610-1 to XZ), the Xiangya Hospital Funds for Talent Introduction (to XZ), and the Hunan Provincial Natural Science Foundation of China (Grant No. 14JJ7008 to XZ).

X.Z. conceived the concept, collected pertinent references, designed and wrote the manuscript, and trained Y.L, X.H.Z, and Y.M regarding statistical significance, biological significance, and systems biology. Y.L. and Y.M participated in the collection of references, discussion and modification of manuscript. X.H.Z participated in revision of the manuscript. All authors approved the final manuscript.

The authors declare that there is no conflict of interests regarding the publication of this article.