Home - Secondary structure Comparative modelling Contacts Threading - search - FTP - Mirrors - Bottom Explain scores

EVA sec results for: SSpro2

Organisation of method specific results

AVERAGES : averages over all weeks
EACH MONTH : average over one month note: detail lists all proteins of that month
FTP RDB-table : flat file with all results note: may be VERY large!!

METHOD about:	../../doc/explain_methods.html#PX_about_sspro2
METHOD url:	SSpro2
METHOD quote:	G Pollastri, D Przybylski, B Rost, P Baldi:: Improving the prediction of protein secondary structure in three and eight classes using recurrent neural networks and profiles. Proteins, 47, 228-335, 2002
TYPE :	sec
NPROT :	257
NWEEKS :	48
AVE_Q3 :	74.3
AVE_SOV:	69.0
NOTE 1 :	here ALL results obtained for sspro2 are given, i.e. NOT those for a common subset!
NOTE 2 :	all proteins evaluated had no SIGNIFICANT sequence identity to any protein of previously known structure!

Averages over all proteins

type	date	sots	id	Q3	Q3H%o	Q3H%p	Q3E%o	Q3E%p	Q3L%o	Q3L%p	sov	sovH	sovE	sovL	BAD	info	infoO	infoP	corrH	corrE	corrL	class	contDH	contDE
sec	2004_04_28	dssp	psum	74.3	81	78	61	72	75	71	69.0	72.8	71.3	66.6	3.0	0.34	0.35	0.34	0.67	0.58	0.53	8171.2	7.2	6.5
sec	2004_04_28	dssp	sig	11.9	31	31	34	40	14	13	16.6	26.7	27.0	17.0	5.1	0.17	0.17	0.17	0.25	0.28	0.18	8105.3	8.5	8.0
sec	2004_04_28	dssp	errsig	0.74	1	1	2	2	0.89	0.82	1.0	1.6	1.6	1.0	0.31	0.01	0.01	0.01	0.01	0.01	0.01	505.5	0.53	0.50

Averages over each month

Type	Method	Month	Nprot	Average Q3 for month	File with details for month
sec	sspro2	ALL	257	74.3
sec	sspro2	2001_04	1	79.1	old/2001_04_sspro2.html
sec	sspro2	ALL	257	74.3

Notations used (detailed explanation)

psum	percentage over all proteins
sig	standard deviation
errsig	significant difference (to distinguish between two methods)

type	prediction type
date	date when added [yyyy_mm_dd]
sots	standard of truth
id	identifier of protein
Q3	three-state per-residue accuracy (percentage of correctly predicted residues)
Q3H%o	correctly predicted helix residues (percentage of helix observed)
Q3H%p	correctly predicted helix residues (percentage of helix predicted)
Q3E%o	correctly predicted strand residues (percentage of helix observed)
Q3E%p	correctly predicted strand residues (percentage of helix predicted)
Q3L%o	correctly predicted loop residues (percentage of helix observed)
Q3L%p	correctly predicted loop residues (percentage of helix predicted)
sov	per segment accuracy (three states)
sovH	per segment accuracy for helix residues
sovE	per segment accuracy for strand residues
sovL	per segment accuracy for non-regular residues
BAD	percentage of confused HELIX/STRAND residues (helix predicted as strand, strand predicted as helix)
infoO	information index accuracy (percentage of observed)
infoP	information index accuracy (percentage of predicted)
corrH	Matthews correlation index for helix
corrE	Matthews correlation index for strand
corrL	Matthews correlation index for loop
var	variance over all proteins (1/N-1)( score(i) - )*2
sig	sigma (standard deviation) over all proteins (sqrt[var])
errsig	significant differences ( standard dev / sqrt(Nprot) )

Last update: Apr 28, 2004
Contact: rost@columbia.edu
WWW: http://cubic.bioc.columbia.edu/eva

Explain scores - Home - Secondary structure Comparative modelling Contacts Threading - search - FTP - Mirrors - Top