Accuracy Assessment of Alphafold2 Structure Predictions

AF2 (V2.0.1) is used for structure predictions with the required databases downloaded from the AF2 GitHub repository^{3 (link)}. Table 1 summarizes the protein models used in the present work. The AF2 structure models of these proteins are shown in Fig. S1 of the Supplementary Information (SI). All protein sequences can be found in the Appendix of the SI.Table 1

Proteins models used in the present work.

Protein	AA¹	MSA²	pLDDT³	IUPRED2³	RMSF (Å)³	PCC⁶	Slope⁶	Int.⁶
a. LanM	133	1832	83.9 ± 19.1	0.39 ± 0.16	5.8 ± 3.3	− 0.84	− 4.9	113
b. DeHa4	300	1890	96.3 ± 6.9	0.25 ± 0.14	0.9 ± 0.9	− 0.94	− 7.2	103
c. PAS-A Domain	108	1138	81.4 ± 16.3	0.20 ± 0.09	1.0 ± 0.7	− 0.65	− 15.5	97
d. AFP Type III	66	1080	96.4 ± 5.7	0.20 ± 0.07	0.7 ± 0.7	− 0.97	− 8.4	103
e. GNE	722	5273	93.2 ± 11.4	0.21 ± 0.12	3.0 ± 1.1	− 0.75	− 9.6	105
f. PAS-Kinase	1323	8644	52.9 ± 27.5	0.43 ± 0.25	5.0 ± 3.9	− 0.63	− 4.0	77
g. inaZ	1200	2050	88.6 ± 16.5	0.41 ± 0.07	3.8 ± 3.2	− 0.65	− 3.3	101
h. Heterodimer⁴: PAS-A, kinase	108 287	1138 1908	89.5 ± 13.0	0.14 ± 0.10	1.3 ± 0.7	− 0.65	− 11.7	110
i. Homodimer⁵: MtMerR	146 146	1825 1825	89.3 ± 13.9	0.36 ± 0.13	3.8 ± 2.5	− 0.66	− 3.7	103
j. NVJP-1	388	0	43.2 ± 5.3	0.84 ± 0.13	10.2 ± 2.4	− 0.03	− 0.1	44
k. Randomized	237	0	32.4 ± 6.2	0.28 ± 0.19	2.1 ± 1.1	− 0.12	− 0.7	34

¹Number of amino acid residues.

²The MSA hits from the BFD³ (Big Fantastic Database). The MSA hits include those that match the protein partial segments.

³Mean ± SD for per-residue pLDDT, IUPRED2 and RMSF values.

⁴Two chains of the heterodimer are PAS-A (108 AA) and kinase (287 AA) domain sequences, respectively.

⁵Both chains of the homodimer have the same sequence of 146 AA.

⁶The Pearson’s correlation coefficient (PCC) between pLDDT and RMSF scores, the slope and intercepts of the linear fitting between them are also listed; note that as pLDDT and the AF2 scores in this work are anticorrelated, and the PCC values are the negative of those shown in the Figures.

Free full text: Click here

Guo H.B., Perminov A., Bekele S., Kedziora G., Farajollahi S., Varaljay V., Hinkle K., Molinero V., Meister K., Hung C., Dennis P., Kelley-Loughnane N, & Berry R. (2022). AlphaFold2 models indicate that protein sequence determines both structure and dynamics. Scientific Reports, 12, 10696.

Publication 2022

43 63 Amino acid Kinase Protein Protein sequences

Corresponding Organization :

Other organizations : Wright-Patterson Air Force Base, General Dynamics (United States), University of Dayton, University of Utah, University of Alaska Southeast

Top 5 similar protocols

Protocol cited in 3 other protocols

Variable analysis

independent variables

Protein models

dependent variables

AA (number of amino acid residues)
MSA (number of MSA hits from the BFD database)
PLDDT (mean ± SD for per-residue pLDDT scores)
IUPRED2 (mean ± SD for per-residue IUPRED2 scores)
RMSF (mean ± SD for per-residue RMSF values in Å)
PCC (Pearson's correlation coefficient between pLDDT and RMSF scores)
Slope (slope of the linear fitting between pLDDT and RMSF)
Int. (intercept of the linear fitting between pLDDT and RMSF)

control variables

AF2 (V2.0.1) protocol used for structure predictions
Required databases downloaded from the AF2 GitHub repository

Annotations

Based on most similar protocols

Etiam vel ipsum. Morbi facilisis vestibulum nisl. Praesent cursus laoreet felis. Integer adipiscing pretium orci. Nulla facilisi. Quisque posuere bibendum purus. Nulla quam mauris, cursus eget, convallis ac, molestie non, enim. Aliquam congue. Quisque sagittis nonummy sapien. Proin molestie sem vitae urna. Maecenas lorem.

As authors may omit details in methods from publication, our AI will look for missing critical information across the 5 most similar protocols.

About PubCompare

Our mission is to provide scientists with the largest repository of trustworthy protocols and intelligent analytical tools, thereby offering them extensive information to design robust protocols aimed at minimizing the risk of failures.

We believe that the most crucial aspect is to grant scientists access to a wide range of reliable sources and new useful tools that surpass human capabilities.

However, we trust in allowing scientists to determine how to construct their own protocols based on this information, as they are the experts in their field.

Ready to get started?

Revolutionizing how scientists
search and build protocols!