This is the initial working version of superdelta() that does not depend on the multtest package.
Superdelta is a robust procedure of gene expression normalization and differential expressed gene detection.
The basic idea of superdelta is as following:
1) For each gene, take the difference (called "delta") of its expressions with each of the other genes;
2) Construct test statistics from these deltas rather than the original expressions;
3) Pursue a robust asymptotic unbiased estimator of the test statistic (an *oracle* statistic);
4) As a bonus, this procedure will return an optimal pairing gene to each gene, which also provides
   information to the underlying biological feature and interaction of genes.

In addition to being tested thoroughly, we hope to implement the following methods in the future, to expand superdelta to be a
comprehensive pipeline that can treat various type of gene expression data (microarray, sequencing, and others):

1. Paired t-test (essentially a one-sample t-test).
2. ANOVA F-test.
3. Possibly a multi-way F-test.
4. Regression t-test.
*5. Further performance tuning by using float instead of double computing in C functions.