An improved algorithm for the maximal information coefficient and its application

The maximal information coefficient (MIC) captures both linear and nonlinear correlations between variable pairs. In this paper, we proposed the BackMIC algorithm for MIC estimation. The BackMIC algorithm adds a searching back process on the equipartitioned axis to obtain a better grid partition than the original implementation algorithm ApproxMaxMI. And similar to the ChiMIC algorithm, it terminates the grid search process by the χ ²-test instead of the maximum number of bins B(n, α). Results on simulated data show that the BackMIC algorithm maintains the generality of MIC, and gives more reasonable grid partition and MIC values for independent and dependent variable pairs under comparable running times. Moreover, it is robust under different α in B(n, α). MIC calculated by the BackMIC algorithm reveals an improvement in statistical power and equitability. We applied (1-MIC) as the distance measurement in the K-means algorithm to perform a clustering of the cancer/normal samples. The results on four cancer datasets demonstrated that the MIC values calculated by the BackMIC algorithm can obtain better clustering results, indicating the correlations between samples measured by the BackMIC algorithm were more credible than those measured by other algorithms.

Density distribution of gene expression profiles and evaluation of using maximal information coefficient to identify differentially expressed genes

The hypothesis of data probability density distributions has many effects on the design of a new statistical method. Based on the analysis of a group of real gene expression profiles, this study reveal that the primary density distributions of the real profiles are normal/log-normal and t distributions, accounting for 80% and 19% respectively. According to these distributions, we generated a series of simulation data to make a more comprehensive assessment for a novel statistical method, maximal information coefficient (MIC). The results show that MIC is not only in the top tier in the overall performance of identifying differentially expressed genes, but also exhibits a better adaptability and an excellent noise immunity in comparison with the existing methods.

Maximal information coefficient applied to differentially expressed genes identification: A feasibility study

BACKGROUND:

The main obstacle encountered in microarray technology is how to mine the valuable information under the profiles and study the genes function.

OBJECTIVE:

Maximal information coefficient (MIC) is a novel, non-parametric statistic that has been successfully applied to genome-wide association studies and differentially gene and miRNA expression analysis. However, the data used in these applications are not gold standard but real data.

METHODS:

Therefore, this study attempts to test the feasibility of MIC for differentially expressed gene identification with simulation data.

RESULTS:

Our experiments indicate that, MIC perfermance is better than Limma always, which is almost the same level of SAM, ROTS or DESeq2. However, the count of AUC < 0.5 of MIC is significantly smaller than the three methods, and MIC does not exhibit an abnormal phenomenon in which the AUC increases as the noise increases.

CONCLUSIONS:

Compared to the existing methods, our experiments show that MIC is not only in the first tier in identifying differentially expressed genes and noise immunity, but also shows better robustness and stronger data/environment adaptability.