CONSTRUCTION OF A DATABASE OF CO -OCCURRING eMOTIFS BASED ON CONDITIONAL PROBABILITIES

Classification of a newly discovered protein into a family of proteins enables the determination of its function. The eMOTIF system identifies conserved modular domains that confer functionality or structure to proteins and allows classification of proteins into families based on the conserved domains a protein contains. A program called multeeMOTIF has been developed which analyzes eMOTIFS and determines the conditional probabilities of their occurrences to find pairs of eMOTIFS that occur together a percentage of the time. The proteins that match one eMOTIF compose a large super-family of proteins and proteins that match each additional eMOTIF compose smaller and smaller sub-families. Based on how many and which eMOTIFS an unknown protein matches, it can be assigned to the appropriate sub-family. The more eMOTIFS matched, the more specific the family assignment will be. Those pairs of eMOTIFS that always occur together or occur together a high percentage of the time are observed to be from alignments of the same protein functionality 83% of the time. This may allow assignment of function to alignments of unknown function and consequently proteins belonging to such alignments.