Detecting coevolution without phylogenetic trees? Tree-ignorant metrics of coevolution perform as well as tree-aware metrics

Table 3 Tree-ignorant methods compared on myoglobin and myosin.

(A) Tetrapod Myoglobin
	MI		NMI		RMI		SCA		MIP
	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	χ ²	p – value
(i,i+3)	2	51	9	44	0	53	10	43	11	42	17.98	1.24 × 10^-3
(i,i+4)	32	21	41	12	1	52	36	17	18	35	79.28	2.48 × 10 ^-16
(B) Randomized Tetrapod Myoglobin
	MI		NMI		RMI		SCA		MIP
	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	χ ²	p – value
(i,i+3)	4	49	0	53	0	53	0	53	0	53	16.25	2.71 × 10 ^-3
(i,i+4)	0	53	0	53	0	53	0	53	1	52	4.02	4.04 × 10^-1
(C) Chordate Myosin
	MI		NMI		RMI		SCA		MIP
	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	χ ²	p – value
(i,i+3)	51	2	46	7	38	15	40	13	46	7	14.83	5.08 × 10 ^-3
(i,i+4)	52	1	44	9	26	27	44	9	43	10	41.30	2.33 × 10 ^-8
(D) Randomized Chordate Myosin
	MI		NMI		RMI		SCA		MIP
	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	p ≤ α	p > α	χ ²	p – value
(i,i+3)	5	48	18	35	4	49	2	51	27	26	53.30	7.38 × 10 ^-11
(i,i+4)	0	53	0	53	0	53	0	53	0	53	n/a	n/a

χ² goodness-of-fit tests comparing the performance of tree-ignorant methods for detecting alpha helix periodicity at (i, i + 3) and (i, i + 4). Bolded rows highlight statistically significant χ² results, indicating a difference between the methods. Bold counts in these rows highlight which method achieved the highest ratio of significant to insignificant scores. Counts are calculated using the empirically determined optimal SCA cutoff for each positive control data point. SCA cutoff is 0.9 for A-B row 1, 0.8 for A-B row 2, 0.4 for C-D row 1, and 0.6 for C-D row 2. α = 0.01.

ISSN: 2730-7182