Table 5. Frequencies of three different relationships between two neighboring parnters in chimeric, trimeric and tetrameric RNAs.

Database	Relationship	N	Mean	Median(Q2)	Maxima(Q4)	Q1	Q3	Sstdv	Mad
mRNA in NCBI	Exact junction	157
	Overlapped	736	6.5	3	606	2	6	29.52	2.97
	Gap (insert)	373	132.4	49	1303	13	167.5	198.79	65.23
EST in NCBI	Exact junction	2520
	Overlapped	23252	6.4	5	190	3	7	7.7	2.97
	Gap (insert)	8998	29.5	21	307	8	38	31.4	20.76

Note: Putative chimeric, trimeric and tetrameric RNAs were identified from mRNA and EST collections of the NCBI database by our simple computer code. “Exact junction” means that the two partner mRNAs join directly. “Overlapped” means that the two partner mRNAs have at least 1 nucleotide (nt) overlapped. “Gap (insert)” means that there is at least 1-nt or an unmatchable sequence inserted between the two partner RNAs. The length of the overlapped or inserted sequences is short in most cases and is not normally distributed. Therefore, the mean, midian (Q2) and maximal lengthes in number of nt are calculated, besides the length at the 25%th (Q1) and 75% (Q3). Mad(X)  = 1.4826 * median (|Xi-median(X)|). * When the distribution is so different from normal distribution, we usually compare “mean” to “median” and “standard deviation” to “mad”. If the difference between mean and median, STDEV and MAD are huge, the distribution is far from normal distribution.