GC Content and Common Hexamers

Figure 4.6 summarizes GC content in training sets, among validation sequences from Glomus and Medicago spp., and in test libraries. Plant sequences have lower GC than in fungi, which in turn have lower GC than rhizobacteria. The same is true for validation sequences, though the separation is not as great as in training sets. Curiously, this pattern is reversed for axenic plant and fungal test sequences. The Mt Long library has GC content comparable to validation sequences sampled from M. truncatula, intermediate between GC in plant and fungal training sets. Libraries from Glomus intraradices have about 10% lower GC than the fungal training set, and resemble more closely the plant training set. Overlap in GC content with the fungal training set is greatest for the Lammers library, less so for the Harrison library, and not at all for the Sawaki library.

Figure 4.6: GC content of training sets, validation sequences from Glomus and Medicago spp., and test libraries. Error bars indicate standard deviation of the mean.

If we compare GC content (Figure 4.6) with abundant hexamers (Table 4.4 and 4.5), we note that the two comparative measures of composition are related within a set of sequences. Test sequences contain AT-rich common hexamers, and GC residues appear in the Long plant library more often than in the fungal libraries (Table 4.5). This pattern is the opposite of what was seen in training sets (Table 4.4), where common plant hexamers are dominated by T residues, and common fungal hexamers are less biased in composition.

Table 4.4: Common hexamers among training sequences. For each taxon, the ten most abundant hexamers (by rank) in each training set, with percentage of all hexamers represented by that hexamer, are shown.

RANK	FUNGI		PLANTS		RHIZOBACTERIA
1	0.158	CAAGAA	0.191	TTTTTT	0.143	CGCCGC
2	0.152	TGGTAT	0.177	TATTTT	0.142	GGCGGC
3	0.151	TCAAGG	0.168	TTTTAT	0.142	GCGGCG
4	0.148	GTCAAG	0.166	TTATTT	0.142	GCCGGC
5	0.147	TCAAGA	0.162	ATTTTT	0.137	CGGCGC
6	0.146	CAAGGA	0.161	TGTTTT	0.133	CGGCGA
7	0.128	CTGGTA	0.159	TTTGTT	0.130	TCGCCG
8	0.127	ATCAAG	0.157	TTTATT	0.129	TCGGCG
9	0.126	GCTGGT	0.155	TTGTTT	0.127	GCGCCG
10	0.126	CGTCAA	0.154	TTAATT	0.127	CCGGCG

Table 4.5: Ten most common hexamers among test sequence libraries. Shown for each library are the ten most abundant hexamers (by rank), with percentage of all hexamers represented by that hexamer.

RANK	MT LONG		GI HARRISON		GI LAMMERS		GI SAWAKI
1	0.173	GAAGAA	0.247	AAAGAA	0.225	TTATTA	0.263	AAGAAA
2	0.149	AAGAAG	0.230	AAAAGA	0.221	TTTATT	0.246	AAAGAA
3	0.143	AAGAAA	0.215	AAAAAT	0.221	ATTATT	0.227	GAAAAA
4	0.126	TTCTTC	0.212	AAGAAA	0.212	ATTTTT	0.223	AAAAAT
5	0.122	TGAAGA	0.207	AAAGAT	0.192	TATTAT	0.222	AAAAGA
6	0.122	TTGTTG	0.204	CAAAAA	0.183	TTATTT	0.215	AATAAA
7	0.120	AAAGAA	0.202	AAAAAG	0.183	TATTTT	0.205	AAAATT
8	0.116	AGAAGA	0.195	TGATGA	0.177	AATTTT	0.200	AAATTA
9	0.115	TGATGA	0.191	AAAATT	0.172	TTTTAT	0.197	ATTATT
10	0.113	TTTGTT	0.189	ACAAAA	0.172	AATAAT	0.195	TTTATT