Dear Gerard and all,
I'd expect "zero + noise" to be sometimes negative, sometimes positive. What I get is almost always negative and almost never positive. Am I the only one observing this? Here a few random examples, I could post many more. CCanom is the third to last column. That does not behave as "zero plus noise", the noise is always negative and somewhat resolution dependent.
Just adding evidence in case there is something to examined with the math. All of these crystals were well behaved, even if less than stellar in quality, from two different proteins. In both cases the anomalous signal is only from Sulphur (and the occasional phosphorus) and no attempt was made to measure it or to adapt data collection to it.
The numbers are smaller and noisier than those of Jakob, but still.
From a practical point of view I have gotten used to interpreting them as noise, and they don't seem to do any harm.
SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
RESOLUTION NUMBER OF REFLECTIONS COMPLETENESS R-FACTOR R-FACTOR COMPARED I/SIGMA R-meas Rmrgd-F Anomal SigAno Nano
LIMIT OBSERVED UNIQUE POSSIBLE OF DATA observed expected Corr
9.73 2313 502 752 66.8% 4.2% 4.3% 2258 28.75 4.7% 3.2% -3% 0.794 227
7.17 3614 803 1057 76.0% 5.9% 5.8% 3499 21.10 6.6% 5.4% -10% 0.793 367
5.94 4652 1038 1327 78.2% 10.9% 10.9% 4487 12.21 12.1% 11.7% -8% 0.768 435
5.18 5514 1201 1528 78.6% 14.2% 15.1% 5326 9.73 15.7% 16.0% -9% 0.706 509
4.65 6218 1372 1730 79.3% 13.2% 13.1% 5998 10.61 14.6% 13.4% -5% 0.781 573
4.26 6837 1502 1883 79.8% 14.7% 14.5% 6593 9.74 16.2% 14.9% -6% 0.795 642
3.95 7414 1647 2043 80.6% 20.2% 20.4% 7140 7.21 22.4% 22.7% -2% 0.766 702
3.70 8050 1784 2192 81.4% 30.1% 30.7% 7761 4.88 33.3% 33.1% -2% 0.778 758
3.50 8089 1818 2297 79.1% 41.7% 42.9% 7786 3.47 46.3% 45.8% -1% 0.770 772
total 52701 11667 14809 78.8% 15.8% 16.0% 50848 9.59 17.5% 19.0% -4% 0.771 4985
SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
RESOLUTION NUMBER OF REFLECTIONS COMPLETENESS R-FACTOR R-FACTOR COMPARED I/SIGMA R-meas CC(1/2) Anomal SigAno Nano
LIMIT OBSERVED UNIQUE POSSIBLE OF DATA observed expected Corr
6.56 32567 4913 4990 98.5% 3.5% 3.3% 32563 49.81 3.8% 99.9* -13 0.808 4390
4.66 60378 8699 8759 99.3% 4.1% 3.9% 60372 41.67 4.4% 99.9* -18 0.760 8177
3.81 75156 11155 11232 99.3% 4.6% 4.2% 75140 36.52 4.9% 99.8* -15 0.776 10435
3.30 90133 13188 13263 99.4% 6.8% 6.5% 90097 24.36 7.4% 99.7* -10 0.779 12388
2.95 105864 14947 15003 99.6% 13.2% 13.5% 105839 13.21 14.3% 99.3* -4 0.758 14332
2.70 112854 16476 16562 99.5% 25.9% 27.7% 112800 6.85 28.1% 97.2* -1 0.736 15605
2.50 126773 17927 17988 99.7% 53.5% 58.2% 126742 3.59 57.7% 90.9* -1 0.701 17233
2.34 120722 18736 19313 97.0% 96.1% 105.2% 120569 1.80 104.5% 71.8* 0 0.674 17445
2.20 76956 15530 20523 75.7% 147.1% 160.3% 76482 0.90 164.6% 40.5* 1 0.639 12941
total 801403 121571 127633 95.3% 7.7% 7.7% 800604 14.46 8.3% 99.9* -4 0.726 112946
SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= -3.0 AS FUNCTION OF RESOLUTION
RESOLUTION NUMBER OF REFLECTIONS COMPLETENESS R-FACTOR R-FACTOR COMPARED I/SIGMA R-meas Rmrgd-F Anomal SigAno Nano
LIMIT OBSERVED UNIQUE POSSIBLE OF DATA observed expected Corr
10.84 5434 1095 1125 97.3% 4.5% 4.5% 5434 30.21 5.0% 2.8% -18% 0.752 834
7.76 9571 1862 1877 99.2% 5.5% 5.2% 9571 25.22 6.1% 3.8% -21% 0.748 1572
6.37 12736 2393 2406 99.5% 12.0% 11.7% 12736 12.24 13.3% 11.1% -10% 0.775 2170
5.52 14954 2788 2803 99.5% 20.7% 20.8% 14954 7.77 22.9% 18.8% -4% 0.774 2581
4.95 16875 3140 3152 99.6% 24.9% 25.2% 16875 6.56 27.5% 22.1% -6% 0.767 2914
4.52 18733 3483 3496 99.6% 25.0% 24.9% 18733 6.40 27.6% 23.5% -6% 0.774 3256
4.19 20269 3766 3783 99.6% 34.5% 34.8% 20269 4.68 38.1% 34.3% -5% 0.752 3534
3.92 21857 4043 4057 99.7% 52.3% 54.3% 21857 3.01 57.8% 55.2% -3% 0.719 3815
3.70 22450 4232 4317 98.0% 93.6% 96.9% 22401 1.73 103.7% 95.8% 0% 0.704 3890
total 142879 26802 27016 99.2% 18.3% 18.5% 142830 7.87 20.3% 25.1% -5% 0.748 24566
SUBSET OF INTENSITY DATA WITH SIGNAL/NOISE >= 9.0 AS FUNCTION OF RESOLUTION
RESOLUTION NUMBER OF REFLECTIONS COMPLETENESS R-FACTOR R-FACTOR COMPARED I/SIGMA R-meas Rmrgd-F Anomal SigAno Nano
LIMIT OBSERVED UNIQUE POSSIBLE OF DATA observed expected Corr
7.73 5802 1326 1447 91.6% 3.1% 4.4% 5725 30.19 3.5% 1.7% -32% 0.496 490
5.61 9064 2012 2258 89.1% 4.4% 4.7% 8932 27.26 4.9% 2.8% -22% 0.664 806
4.62 11819 2518 2827 89.1% 4.7% 4.8% 11673 27.33 5.2% 3.0% -22% 0.656 1076
4.02 13451 2823 3304 85.4% 5.2% 5.0% 13317 26.46 5.8% 3.4% -22% 0.711 1230
3.60 13080 2677 3747 71.4% 7.0% 6.0% 12981 22.89 7.9% 4.5% -14% 0.849 1243
3.29 12276 2229 4085 54.6% 8.7% 8.0% 12229 19.96 9.6% 4.8% -5% 0.812 1198
3.05 9642 1476 4428 33.3% 11.5% 11.8% 9635 17.46 12.5% 5.6% -7% 0.754 969
2.86 7372 953 4744 20.1% 15.3% 16.5% 7371 15.46 16.4% 6.3% -8% 0.707 751
2.70 4155 503 5055 10.0% 19.0% 20.5% 4155 13.98 20.3% 6.3% -3% 0.723 420
total 86661 16517 31895 51.8% 5.8% 6.0% 86018 23.71 6.5% 3.6% -14% 0.728 8183
Cheers,
Jose.
================================
Jose Antonio Cuesta-Seijo, PhD
Carlsberg Laboratory
Gamle Carlsberg Vej 10
DK-1799 Copenhagen V
Denmark
Tlf +45 3327 5332
Email [log in to unmask]
================================
-----Original Message-----
From: Gerard Bricogne [mailto:[log in to unmask]]
Sent: Friday, July 17, 2015 10:31 AM
To: Seijo, Jose A. Cuesta
Cc: [log in to unmask]
Subject: Re: [ccp4bb] Negative CCanom
Dear Jose,
What you write gives the impression that you expect an absence of anomalous signal to give rise to a zero CCanom (but correct me if I am wrong), negative values being some sort of round-off error. A zero signal, however, would mean that the differences are pure noise, in which case their CC can be negative as well as positive, but with rather small absolute values, meaning that there is no *significant* correlation.
What was odd in Jacob Keller's case was that CCanom had absolute values large enough to indicate a significant *anti-correlation*.
I hope I didn't misinterpret what you were saying.
With best wishes,
Gerard.
--
On Fri, Jul 17, 2015 at 07:47:42AM +0000, Seijo, Jose A. Cuesta wrote:
> Going back to the original discussion on XDS producing negative CCanom, I am not sure that this is the result of some crystal artifact. I just checked many datasets from the last years and in all cases virtually for every every resolution shell, XDS reports a negative CC anom.
> Please note that this is not the case in cases where an anomalous signal is present, in those cases the signal drops to zero smoothly and occasionally at high resolution a CC anom of -1% pops up.
> I am talking of cases where there is no significant anomalous signal. Then it is reported negative and often "higher" (lower) than -10%.
> I must say that my typical data collection and data processing involves pushing a bit both with resolution and with radiation dose, so maybe this is partly the reason.
> IMO this is XDS trying to make the best of a signal that isn't there and turning zero + noise into zero + "negative" noise somehow.
> Just again, if there is a significant anomalous signal, then CCanom behaves as one would expect, positive at low res, decaying at medium res and 0+-noise at high res.
>
> Cheers,
>
> Jose.
>
> ================================
> Jose Antonio Cuesta-Seijo, PhD
> Carlsberg Laboratory
> Gamle Carlsberg Vej 10
> DK-1799 Copenhagen V
> Denmark
>
> Tlf +45 3327 5332
> Email [log in to unmask]
> ================================
>
>
> -----Original Message-----
> From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of
> Kay Diederichs
> Sent: Friday, July 17, 2015 8:00 AM
> To: [log in to unmask]
> Subject: Re: [ccp4bb] Negative CCanom
>
> Am 17. Juli 2015 00:00:11 MESZ, schrieb "Keller, Jacob" <[log in to unmask]>:
> >>Jacob's case is twinning in P3(2)12 making the data appear as
> >>P6(2)22,
> >so it is indeed a rotation by 180°.
> >
> >Yes, this all fits together nicely. If I understand correctly, this
> >would make my crystals a blend of mero- and pseudo-mero-hedral, so
> >involving I's and F's, no?
>
> No, it's just merohedral - the true and the apparent spacegroup belong to the same pointgroup.
> Even if it were pseudo-merohedral it would be a superposition of reflections, with addition of their intensities.
>
> HTH,
>
> Kay
>
> >
> >
> >
> >JPK
>
> --
> Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail gesendet.
--
===============================================================
* *
* Gerard Bricogne [log in to unmask] *
* *
* Global Phasing Ltd. *
* Sheraton House, Castle Park Tel: +44-(0)1223-353033 *
* Cambridge CB3 0AX, UK Fax: +44-(0)1223-366889 *
* *
===============================================================
|