advertisement: compare things at compare-stuff.com!

Mainly-$\beta$ architectures

Within the mainly-$\beta$ class there are architectural differences in pattern composition which help explain the prediction results given in Section 3.4.2. Highly significant $\chi _t^2$ scores are seen for Cys-containing duplets between ribbon and non-ribbon architectures (Table 3.7). Disulphide bonds are important in ribbon architectures; the cysteine knot topology (CATH code 2.10.90) has a very well conserved pattern of Cys residues (see also Chapter 4). The overabundance of Cys in the ribbon architecture is therefore easily explained.

Table 3.7: Chi-squared analysis of (i,i+3) duplet usage between the ribbon and non-ribbon beta architectures. Values of $\chi_t^2>23$ are significant at the level of $P'<0.001$. Five duplets were excluded as they did not occur in the dataset.

			Observed - Expected ($O-E$)
pattern		$\chi _t^2$	Other	Ribbon
C	. .	C	300.0	-13.3	13.3
Y	. .	C	68.5	-3.7	3.7
C	. .	G	66.7	-7.1	7.1
C	. .	K	37.0	-2.7	2.7
C	. .	F	31.4	-0.9	0.9
M	. .	C	31.4	-0.9	0.9
H	. .	C	29.4	-1.8	1.8
E	. .	C	27.5	-2.7	2.7
C	. .	T	27.5	-2.7	2.7
N	. .	C	24.2	-2.6	2.6
A	. .	C	24.2	-2.6	2.6

The $\chi _t^2$ analysis of duplet frequencies across the three main non-ribbon architectures (sandwich, barrel and distorted sandwich) does not give such a clear picture (in agreement with the poorer predictions in Section 3.4.2). None of the duplets has significant $\chi _t^2$ results at the level of $P'<0.001$ (Table 3.8 shows the top 10 duplets ranked by $\chi _t^2$). Nevertheless, there appears to be a trend towards Gly containing duplets (Gly-X-X-Tyr, Asp-X-X-Gly and to a lesser extent Gly-X-X-Asp) in the barrel architecture. This suggests that flexibility of Gly is employed to create the barrel architecture, perhaps in $\beta$ turns and hairpins. In Figure 3.4(c), duplets with $\chi_t^2>13$ are shown in red on the porin structure[Cowan et al., 1992], domain 1pho00. This is a very large barrel, although not the most typical, perhaps. No under-abundant duplets are present in this structure (they would be shown in blue). In this figure, the three Gly-containing duplets appear to be found in or near turns. Quantitative $\chi_d^2$ analysis of domains from these three architectures shows that Gly-X-X-Tyr is found mainly in strand ($P=0.001$). Asp-X-X-Gly is very significantly over-abundant in turns ($P<0.001$). The secondary structural preference (for turn) of Gly-X-X-Asp is not significant.

Table 3.8: Chi-squared analysis of (i,i+3) duplet usage between the sandwich, barrel and distorted sandwich architectures. Values of $\chi_t^2>26$ are significant at the level of $P'<0.001$ (i.e. none). 17 duplets were excluded as they did not occur in the dataset.

			Observed - Expected ($O-E$)
pattern		$\chi _t^2$	Sand.	Barrel	Dist. Sand.
G	. .	Y	21.2	-8.8	8.9	-0.1
D	. .	G	18.5	-11.4	10.0	1.4
G	. .	C	14.6	-2.3	-1.4	3.7
A	. .	C	14.5	-3.0	-0.0	3.1
C	. .	Q	14.4	-1.8	-0.6	2.4
G	. .	D	13.6	-6.9	6.5	0.3
T	. .	K	9.1	-3.4	-0.5	3.9
A	. .	Y	8.8	-2.6	5.3	-2.7
D	. .	R	7.9	-2.7	2.7	-0.0
P	. .	V	7.4	6.5	-4.2	-2.3