Ten difficult cases

Table I: Results – Pos. – where was the structure found in the list of hits; N^A – number of aligned residues between matched structures; r.m.s.d. – root mean square deviation. Pdb ID if present reflect the structure that was found instead of structure mentioned in second column (might be connected with information about sequence identity with the structure from second column). Pride is not covered since it didn’t succeed in any single case.

Query	Non-trivial similarity	Dali Pos., N^A, r.m.s.d. (Å)	Vast Pos., N^A, r.m.s.d. (Å)	Matras position	Dejavu Pos., N^A, r.m.s.d. (Å)	SSM Pos., Match	Top Pos., N^A	Tops Pos. score	Lock Pos., N^A, r.m.s.d. (Å)	Topscan Pos., score	CE Pos., N^A, r.m.s.d. (Å)
1fxi:a	1ubq	11., 64, 3,2	6., 51, 2,3	1ndd(57%) 23	16., 52, 1,794	45., 83 %	No	Yes, score 2	3., 50, 1,774	no	no
1ten:	3hhr:b	10., 86, 2,1	101., 60, 1,6	1axiB 85	no	No	no	Yes Score3	no	19.	51., 87, 1,9
3hla:b	2rhe	13, 86, 2,1	no	No	no	No	8fab, 31., 74	no	7., 53, 1,769	no	Far from the beg, 84, 3,4
2aza:a	1paz	16., 124, 2,6	22., 69, 2,2	1bqk(66%) 15	21.,74, 1,750	No	no	no	6.,67, 1,769	Yes , Score51	Far, 85, 2,9
1cew:I	1mol:a	4., 81, 2,3	12., 66, 2,0	3	1., 74, 1,655	No	Yes 2, 63	3., score3	3., 57, 1,752	no	7., 74, 2,1
1cid	2rhe	12., 93, 2,8	No with subset	No	No	No	no	no	17., 49, 1,864	Yes? Score 37,857	f., 96, 3,4
1crl	1ede	18, 210, 3,5	30., 186, 3,4	1b6q 23	No	No	no	NO???	4., 114, 2,082	57., score 29,333 ??	f., 213, 4.1
2sim	1nsb:a	9., 3,8, 287	7., 291, 3,8	1nsc 7	41., 130, 2,170	No	no	Yes Score 4	2., 208, 1.996	9., score 55,0	f., 241, 3,6
1bge:b	2gmf:a	no	8., 68, 2,1	18	No	No	99., 59	no	9., 74, 2,00	no	>200, 107, 4,1
1tie	4fgf	10., 109, 2,9	24., 76, 1,9	1bfg 8	26., 93, 1,636	No	11., 76	no	3., 97, 1,577	no	28., 128, 3,2
1g61	1bwd	5., 182, 3,6	5, 161, 3,9	5	8., 116, 2,137	No	Yes, 97	1.,4jdw 19	no	1., 4jdw	2., 142, 3,6

Notes to results:

Dali found to 3hla:b only 8fab, which is structural representative for 2rhe a values in the table are for 8fab.
Values for top to 3hla:b are to 8fab.

Vast values for aza:2 are actually for structure 1pmy and lock values are for structure 1pzc

Tops found 1ede as a result for 1crl on 15^th position between irrelevant results (score 45)

Results (values) for 2aza with Dali server are for 1bqk structure, which has 65%identity with 1paz and is structure representative in FSSP for 1paz.

Dali results to 1cid are for structure 8fab:A. Dali results to 2sim are for structure 2qwc, which is only 30 identical to 1nsb.

Dali didn’t find the result for 1bge:b, although it found it three years ago.
CE found the result for 1bge:b only in the whole database and not in the representatives, but the 2gmf is in the set of representatives.
Lock values for 1tie are to structure 2fbf, Vast values are to 1bff and top values to 1bfg, what is everything similar protein.

Vast did not find result for 1cid in subset of PDB database, but found it in all PDB database on position 267, 85 aligned residues with rmsd 1,6

Table II: Processing times - given values refer to period of time among sending the query and obtaining the results. I have used pre-computed results in the case of CE server, therefore I don’t have any values for CE processing times (more in paper).

	Dali	Vast	Dejavu	Matras	SSM	Top	Tops	Lock	Topscan
1fxi:a	3 min	10 min	73 min	6 min	Sec	65 min	2 min	7 min	sec
1ten:	9 min	10 min	88 min	18 min	1 min	62 min	2 min	12 min	Sec
3hla:b	8 min	8 min	17 min	34 min	1 min	7 min	6 min	10 min	Sec
2aza:a	ND	15 min	28 min	27 min	2 min	10 min	3min	8 min	sec
1cew:I	5min	10 min	94 min	15 min	Sec	8 min	3 min	10 min	Sec
1cid:	ND	8 min	94 min	17 min	1 min	10 min	5 min	11 min	Sec
1g61	6 min	27 min	38 min	23	5 min	21 min	12 min	24 min	sec
1crl	12 min	85 min	170 min	34 min	4 min	16 min	7 min	26 min	1 min
2sim	6 min	ND	115 min	53 min	2 min	10 min	20 min	5O min	40 sec
1bge:b	4 min	6 min	28 min	12 min	Sec	28 min	1 min	7 min	sec
1tie	150 min	10 min	48 min	30 min	2 min	15 min	6 min	12 min	sec

Table III: Summary of results

method	Dali	Vast	Dejavu	Matras	Pride	SSM	Top	Tops	Topscan	Lock	CE
Correct results	10/11	9/11	6/11	9/11	0/11	1/11	5/11	5/11	6/11	9/11	10/11

Ten difficult cases other way round:

I have tried one of the ten difficult cases (1cew) also other way round: that means I took the structure, which I was supposed to find in first search and used it as a query. I looked if I can find the previous query among the results and I checked also the given values for the quality of match (scores, rmsd’s)

I found out that the results and even the scores were as expected with for all the servers except of Vast,Top, Topscan, Dejavu and Pride. Vast, Top and Topscan found the structure, but with different scores and Dejavu and Pride failed in this test.

Vast : 1mol as a query found 73 aa overlap and rmsd 1,8 for structure 1 cew, while 1cew as a query found overlap of only 66 amino acids and rmsd 2,0 !

Top: 1cew as a query: values 4,74 structural diversity, 63 aligned residues, 18, 7 topological diversity and 5 aligned SSE.
1mol as a query gives 3,54 structural diversity, 64 aligned residues, 14, 5 topological diversity and 4 aligned SSE

Topscan found 1dim (another structure of the same protein as 2sim) with 1nsb:A as the query structure with score 64,15, while 2sim as a query found 1nsb with score 55,0

Table IV: Structures’ summary – this list shows only structures that were used as a query

	Resolution (Å)	lenght	SSE content (%)
1fxi:a	2,2	96	38
1ten:	1,8	90	53
3hla:b	2,6	99	50
2aza:a	1,8	129	49
1cew:I	2,0	108	69
1g61:A	1,3	225	58
1cid:	2,8	177	58
1crl	2,06	534	50
2sim	1,6	381	50
1bge:b	2,2	159	72
1mol:A	1,7	94	71
1tie	2,5	166	43

I mentioned that 1cid has two identical domains in one chain, which fact can biased the results in some of the programs. I used in my test whole chain according to recipe (Fischer et al.,1996).

---

Comments

More about us