DMFold Q9UBK2

DMFold results for Q9UBK2

[Click result.zip to download all results on this page]

Input Sequence in FASTA format

>Q9UBK2 (798 residues)
MAWDMCNQDSESVWSDIECAALVGEDQPLCPDLPELDLSELDVNDLDTDSFLGGLKWCSD
QSEIISNQYNNEPSNIFEKIDEENEANLLAVLTETLDSLPVDEDGLPSFDALTDGDVTTD
NEASPSSMPDGTPPPQEAEEPSLLKKLLLAPANTQLSYNECSGLSTQNHANHNHRIRTNP
AIVKTENSWSNKAKSICQQQKPQRRPCSELLKYLTTNDDPPHTKPTENRNSSRDKCTSKK
KSHTQSQSQHLQAKPTTLSLPLTPESPNDPKGSPFENKTIERTLSVELSGTAGLTPPTTP
PHKANQDNPFRASPKLKSSCKTVVPPPSKKPRYSESSGTQGNNSTKKGPEQSELYAQLSK
SSVLTGGHEERKTKRPSLRLFGDHDYCQSINSKTEILINISQELQDSRQLENKDVSSDWQ
GQICSSTDSDQCYLRETLEASKQVSPCSTRKQLQDQEIRAELNKHFGHPSQAVFDDEADK
TGELRDSDFSNEQFSKLPMFINSGLAMDGLFDDSEDESDKLSYPWDGTQSYSLFNVSPSC
SSFNSPCRDSVSPPKSLFSQRPQRMRSRSRSFSRHRSCSRSPYSRSRSRSPGSRSSSRSC
YYYESSHYRHRTHRNSPLYVRSRSRSPYSRRPRYDSYEEYQHERLKREEYRREYEKRESE
RAKQRERQRQKAIEERRVIYVGKIRPDTTRTELRDRFEVFGEIEECTVNLRDDGDSYGFI
TYRYTCDAFAALENGYTLRRSNETDFELYFCGRKQFFKSNYADLDSNSDDFDPASTKSKY
DSLDFDSLLKEAQRSLRR

Predicted Contact and Distance Map

Contact Map

Distance Map

Top 5 final models from DMFold

Two sets of DMFold models are listed: 'v1' refers to the models built with the 'pLDDT' pre-trained parameters and 'v2' to the models built with the 'pTM' pre-trained parameters. On average, the 'v1' models have a slightly higher pLDDT score than the 'v2' models, and users are suggested to select the models with a higher pLDDT score.

Final models from DMFold with parameters v1

DMFold models

Click to view	Rank	Download	pLDDT score^a
	1	model1.pdb.gz	0.5857
	2	model2.pdb.gz	0.5847
	3	model3.pdb.gz	0.5838
	4	model4.pdb.gz	0.5835
	5	model5.pdb.gz	0.5798

(a)

This set of models (v1) is predicted by DMFold with the 'pLDDT' pre-trained parameters, which is slightly better than the models (v2) predicted by DMFold with the 'pTM' pre-trained parameters in average meaning based on our benchmark results. However, no matter with which version, the model with the higher pLDDT score should be always selected. The model confidence is quantitatified and ranked by predicted LDDT (pLDDT) score, and higher pLDDT score signifies higher model confidence. Usually, pLDDT score>0.7 indicates a correct fold.

Residue-level pLDDT scores and alignment depth

Residue-level pLDDT scores of model1 (v1) and alignment depth of the MSA

Final models from DMFold with parameters v2

DMFold models

Click to view	Rank	Download	pLDDT score^a
	1	model1_v2.pdb.gz	0.5315
	2	model2_v2.pdb.gz	0.5311
	3	model3_v2.pdb.gz	0.5309
	4	model4_v2.pdb.gz	0.5227
	5	model5_v2.pdb.gz	0.5224

(a)

This set of models (v2) is predicted by DMFold with the 'pTM' pre-trained parameters, which is slightly worse than the models (v1) predicted by DMFold with the 'pLDDT' pre-trained parameters in average meaning based on our benchmark results. However, no matter with which version, the model with the higher pLDDT score should be always selected. The model confidence is quantitatified and ranked by predicted LDDT (pLDDT) score, and higher pLDDT score signifies higher model confidence. Usually, pLDDT score>0.7 indicates a correct fold.

Residue-level pLDDT scores and alignment depth

Residue-level pLDDT scores of model1 (v2) and alignment depth of the MSA

Predicted aligned error

The color at position (x, y) indicates DMFold's expected position error at residue x in Angstroms, when the predicted and true structures are aligned on residue y.

PAE

References:
1.	Wei Zheng, Qiqige Wuyun, Yang Li, Chengxin Zhang, P Lydia Freddolino, Yang Zhang. Improving deep learning protein monomer and complex structure prediction using DeepMSA2 with huge metagenomics data. Submitted (2023).