Identification of candidate regulatory sequences in mammalian 3' UTRs by statistical analysis of oligonucleotide distributions

Table 2 Comparison between the results of our computational approach and seed regions of known miRNAs.

Method	7-mers identified	corresponding to known miRNA	expected by chance	P-value
CO	465	74	28.9	5.0 · 10^-14
SA	214	41	13.3	7.9 · 10^-11
CSA	113	19	7.01	6.3 · 10^-5
CO ∩ CSA	59	14	3.66	1.1 · 10^-5
CO ∪ SA	601	94	37.3	4.8 · 10^-17

The rows indicate the computational methods as described in the text: CO = conserved overrepresentation; SA = strand asymmetry; CSA = conserved strand asymmetry (oligos displaying strand asymmetry in both human and mouse); CO∩CSA oligos identified by both CO and CSA;CO∪SA: oligos identified by either conserved overrepresentation or strand asymmetry (human) or both. The columns are (1) Number of oligos identified computationally; (2) Number of these matching the seed region of known human miRNAs; (3) number of such matches expected by chance; (4) number of different known human miRNAs putatively binding the oligos (many 7-oligos occur in the seed region of more than one miRNA) (5) P-value from exact Fisher test, taking into account that there are 16384 possible 7-mers 1017 of which are seed regions of known miRNAs.

ISSN: 1471-2105