Association rule analysis is based on simple arithmetic methods. The analysis is start from a ratio of the number of genes possessing and not possessing cirtain cis-elements in their prometer regions.
Although association rule analysis is simple and effective approach, we should note that this method tends to show false-positive results when the number of user-defined target genes is much smaller than that of reference gene list. In default, reference gene list is the whole available TUs stores in KOME database, which include up to 30,000 genes.
We have considered some solutions to cancel this advantage by adding some options to RiCES.
Users can try association rule analysis with smaller reference genes which are selected from the default reference gene set with certain conditions, such as possession of a known cis-element motif in upstream region. We have prepared several such reference sets, which can be selected easily by users in the form of the starting page.
These specialized references would be useful in analyses of co-existence factors under existence of a certain core cis-element.
The size of reference gene list can be reduced by selecting 'Adjust Reference Size to Target TU list'. When this option is selected, twice as many TUs as user-specified target gene list are randomly selected from the reference gene set (see the bottom section of the figure above), and the result gene set will be used for analysis as reference.
This is an experimental feature, which might be effective to cancel limitation of association rule evaluation mentioned above especially when quite small number of TUs are specified as user's gene list.
Users can also check cis-element motifs obtained by their own experiments or literature search. Such candidates can be input through "Motif List" box in the form.
Last Update: 30 Nov., 2007