"... In PAGE 6: ... Strategy 2 uses all available alignment information (pairwise and multiple alignments) whereas strategy 3 does not use any alignment information in the actual motif discovery process. Table3 summarizes the different stages in the motif discovery process for each strategy.... In PAGE 7: ... We prune the list of motif candidates by removing degenerate motifs based on their Z-score and P-values. This step halves the number of motif candidates (see Table3 ). An overview of the entire processing pipeline is given in Figure 3.... ..."

"... In PAGE 3: ... After applying homology alignment for each cluster, we determined a local motif, which had the highly conserved amino acids. Shown in Table1 are the local motifs of the five clusters, where X means any amino acids. In Table 1, cleavage probability ( cleavage prob ) is defined above, homology % is given by the GeneBee (www.... ..."

"... In PAGE 9: ... So, we reverse complemented each motif and compared the resultant pattern with our motif and recorded their degree of palindromicity, which is defined as the ratio of the characters that match when the pattern is compared with itself after it is reverse complemented. Table2 shows that the motifs found using MOPAC have average to high palindromicity. 5.... ..."

"... In PAGE 10: ... So, we reverse complemented each motif and compared the resultant pattern with our motif and recorded their degree of palindromicity, which is defined as the ratio of the characters that match when the pattern is compared with itself after it is reverse complemented. Table2 shows that the motifs found using MOPAC have average to high palindromicity. 5.... ..."

"... In PAGE 4: ... This results in a natural probability distri- bution for mask elements. 3 RESULTS Table3 summarizes the performance of the experiments. Type denotes the basic (B) or Lamarckian (L) runs.... In PAGE 4: ... Here, low FP scores and high FP scores are preferred. The final 3 columns of Table3 report testing perfor- mance for the natural probability distributions for masks, which were derived for the motifs from the basic GP runs only. The natural masks resulted in higher probabil- ity performance only for the zinc finger families, while other families resulted in lower probabilities.... In PAGE 4: ... The use of conserved groups (Figure 2) during evolution was probably a strong factor in their more frequent occurrence. Table 5 shows some testing comparisons between the best basic solutions from Table3 (B entries) and some mo- tifs evolved in earlier work in [12]. The motifs from this earlier work generally show better performance than the... In PAGE 5: ... A result of this, however, is that the sequence recognition problem is more difficult. Conse- quently, testing scores reported in Table3 were often worse than earlier experiments with respect to true positive recog- nition rates. In other words, the motifs are overly discrimi- natory.... ..."

"... In PAGE 5: ... This results in a natural probability distri- bution for mask elements. 3 RESULTS Table3 summarizes the performance of the experiments. Type denotes the basic (B) or Lamarckian (L) runs.... In PAGE 5: ... Here, low FP scores and high FP scores are preferred. The final 3 columns of Table3 report testing perfor- mance for the natural probability distributions for masks, which were derived for the motifs from the basic GP runs only. The natural masks resulted in higher probabil- ity performance only for the zinc finger families, while other families resulted in lower probabilities.... In PAGE 5: ... The use of conserved groups (Figure 2) during evolution was probably a strong factor in their more frequent occurrence. Table 5 shows some testing comparisons between the best basic solutions from Table3 (B entries) and some mo- tifs evolved in earlier work in [12]. The motifs from this earlier work generally show better performance than the... In PAGE 6: ... A result of this, however, is that the sequence recognition problem is more difficult. Conse- quently, testing scores reported in Table3 were often worse than earlier experiments with respect to true positive recog- nition rates. In other words, the motifs are overly discrimi- natory.... ..."

"... In PAGE 4: ... Convergence was robust and consistent structural features of optimized connection matrices were observed. Figure 3B, Figure 5, Table3 , and Table 4 summarize results obtained from the optimizations. When maximizing func- tional motif number (Figure 5A), we obtained networks that closely resembled real brain networks with respect to their structural and functional motif number, motif diversity (unpublished data), structural motif frequency spectrum, and the specific structural motifs that occurred with significantly increased frequency (Tables 3 and 4).... In PAGE 7: ...75.36 (36.34) z = 8.94 Compare motif ID with those shown in Figure 3 and Table 2. As in Table3 , all networks were optimized for high functional motif number (M =3,N = 30, K = 311, mean and standard deviation for n = 10 exemplars). Optimizations and comparisons of macaque and cat matrices produce similar results (unpublished data).... ..."