Re: [ccp4bb] difference between I over sigma I and I over sd

Protein crystallography Programs for crystallography X-ray detectors
Basic tutorials:
Protein Peptide

CCP4bb navigation

CCP4bb <-- 2008 <-- January 2008 <-- 21 January 2008

Previous message:
Subject: antibody crystallization
From: William Scott wgscott {- at -} CHEMISTRY {- dot -} UCSC {- dot -} EDU
Date: 2008-01-20

Next message:
Subject: Spacegroup choices, reindexing and so on
From: "Winter, G (Graeme)" g {- dot -} winter {- at -} DL {- dot -} AC {- dot -} UK
Date: 2008-01-21

Subject: Re: difference between I over sigma I and I over sd
From: "Winter, G (Graeme)" g {- dot -} winter {- at -} DL {- dot -} AC {- dot -} UK
Date: 2008-01-21

Hi Lu Yongzhi,

It has to be said it is not obvious. However, there are as far as I can
see at least three possible definitions of I/sigma, so here we go:

(1) I/sigma (calculated, corrected perhaps) for individual observations
- sigma based on counting statistics
(2) Mean(I) / Sigma(mean) - that is, including the improvement in the
measurement by including multiple observations
(3) Mean(I) / Spread - that is, the mean I value divided by the spread
(standard deviation?) of the observations - which is different to (1)

Quoting from pdb pages:

Item _reflns_shell.meanI_over_sigI_all

Description

The ratio of the mean of the intensities of all
reflections
in this shell to the mean of the standard uncertainties
of the
intensities of all reflections in this shell.

They do not include a definition for the "sd" version of this.

Reading through a handily found scala log file and comparing to the
harvesting stats, I find that:

meanI_over_sd_all = Mn(I)/sd - (2) above
meanI_over_sigmaI_all = I/sigma - (1) above

Hope this answers your question.

Cheers,

Graeme

________________________________

From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Lu
Yongzhi
Sent: 19 January 2008 09:07
To: CCP4BB@JISCMAIL.AC.UK
Subject: [ccp4bb] difference between I over sigma_I and I over sd

Hi everyone,

When I scale the data using Scala, I don't know differences between
_diffrn_reflns.meanI_over_sigI_all 6.5475
_diffrn_reflns.meanI_over_sd_all 21.3013
in the statistics. I think that both "sigma" and "sd" have the same
meaning--"standard deviation". By the way, in the file header of PDB's,
what are the meanings of the numbers 5 and 1 in the row "HELIX", numbers
0, 1 and -1 in the row "SHEET". And in the row "SHEET", what's the
meaning of the last 6 columns.
HELIX 1 1 THR A 35 ASP A 37 5
3
HELIX 2 2 SER A 41 TYR A 46 1
6
HELIX 3 3 GLU A 68 ALA A 82 1
15 SHEET 1 C 6 GLU A 224 ARG A 232 0

SHEET 2 C 6 ILE A 214 GLU A 221 -1 N CYS A 217 O ARG A 229

SHEET 3 C 6 PHE A 183 VAL A 189 -1 N PHE A 187 O LEU A 218

SHEET 4 C 6 VAL A 302 LEU A 309 -1 O ILE A 306 N LYS A 184

SHEET 5 C 6 LYS A 254 HIS A 261 -1 N ILE A 258 O GLY A 305

SHEET 6 C 6 GLU A 280 PRO A 288 1 O ASP A 283 N VAL A 257

Thanks

Lu Yongzhi

CCP4bb navigation

CCP4bb <-- 2008 <-- January 2008 <-- 21 January 2008

Previous message:
Subject: antibody crystallization
From: William Scott wgscott {- at -} CHEMISTRY {- dot -} UCSC {- dot -} EDU
Date: 2008-01-20

Next message:
Subject: Spacegroup choices, reindexing and so on
From: "Winter, G (Graeme)" g {- dot -} winter {- at -} DL {- dot -} AC {- dot -} UK
Date: 2008-01-21