Probe 2.18 leads to slightly different results than 2.16

[tillea]

Hi, in the Debian packaged version of probe of version 2.16 (git commit 9b198c1) we added an CI test which created a file 6ins.kin which is compared with some checksum. Unfortunately this checksum has changed and thus I've created a diff between the two result files which actually show several differences:

--- /tmp/king-probe-test.2.16.160404+git20200121.9b198c1-3/data/6ins.dotinfo.table.head     2021-10-06 10:37:24.349423219 +0200
+++ 6ins.dotinfo.table.head 2021-10-06 10:35:28.076993097 +0200
@@ -8,33 +8,33 @@

 subgroup: self dots
 atoms selected: 1268
-potential dots: 219619
-potential area: 13726.2 A^2
+potential dots: 218753
+potential area: 13672.1 A^2
   type                 #      %       score score/A^2 x 1000
-  C wide_contact    2716   1.2%      24.9      1.81
-  C close_contact    1733   0.8%      76.0      5.53
-  C small_overlap     576   0.3%     -18.3     -1.34
-  C bad_overlap      21   0.0%      -4.2     -0.31
+  C wide_contact    2748   1.3%      25.0      1.83
+  C close_contact    1755   0.8%      76.7      5.61
+  C small_overlap     605   0.3%     -19.6     -1.43
+  C bad_overlap      22   0.0%      -4.6     -0.34
   C H-bond     281   0.1%       7.9      0.58
-  N wide_contact     550   0.3%       5.7      0.41
-  N close_contact     503   0.2%      23.0      1.67
-  N small_overlap     447   0.2%     -20.6     -1.50
-  N bad_overlap      67   0.0%     -10.6     -0.77
+  N wide_contact     543   0.2%       5.6      0.41
+  N close_contact     499   0.2%      22.7      1.66
+  N small_overlap     444   0.2%     -20.5     -1.50
+  N bad_overlap      67   0.0%     -10.6     -0.78
   N H-bond      61   0.0%       2.1      0.15
-  O wide_contact    1806   0.8%      18.0      1.31
-  O close_contact    1608   0.7%      72.4      5.28
-  O small_overlap    1304   0.6%     -61.9     -4.51
-  O bad_overlap     240   0.1%     -43.1     -3.14
-  O H-bond    2964   1.3%      87.2      6.35
-  S wide_contact     212   0.1%       2.1      0.15
-  S close_contact      82   0.0%       3.3      0.24
-  S small_overlap       3   0.0%      -0.1     -0.00
+  O wide_contact    1800   0.8%      17.8      1.30
+  O close_contact    1603   0.7%      72.2      5.28
+  O small_overlap    1305   0.6%     -62.2     -4.55
+  O bad_overlap     237   0.1%     -42.8     -3.13
+  O H-bond    2939   1.3%      86.4      6.32
+  S wide_contact     209   0.1%       2.0      0.15
+  S close_contact      87   0.0%       3.5      0.26
+  S small_overlap       1   0.0%      -0.0     -0.00
   S H-bond      36   0.0%       1.2      0.09

-     tot contact:     9210   4.2%     225.2     16.41
-     tot overlap:     2658   1.2%    -158.7    -11.56
-     tot  H-bond:     3342   1.5%      98.4      7.17
+     tot contact:     9244   4.2%     225.6     16.50
+     tot overlap:     2681   1.2%    -160.2    -11.72
+     tot  H-bond:     3317   1.5%      97.7      7.14

-       grand tot:    15210   6.9%     164.9     12.01
+       grand tot:    15242   7.0%     163.1     11.93

-contact surface area: 950.6 A^2
+contact surface area: 952.6 A^2

I wonder whether this kind of differences are expected or whether something is wrong. The same is true for the other test data set that is used later in our test script.

Kind regards, Andreas.

[russell-taylor]

The difference is expected and is the result of using the proper radius for C=O carbonyl atoms.

You should be able to reproduce the earlier results by using the -COSCALE0.942857 command-line option to set it back to its original value of 1.65/1.75. Its new value is 1.65/1.7 to match the new radius of these atoms.

The 2.18 results are more correct than the 2.16 results.

[tillea]

Am Wed, Oct 06, 2021 at 06:18:12AM -0700 schrieb Russell Taylor:

The difference is expected and is the result of using the proper radius for C=O carbonyl atoms.

You should be able to reproduce the earlier results by using the -COSCALE0.942857 command-line option to set it back to its original value of 1.65/1.75. Its new value is 1.65/1.7 to match the new radius of these atoms.

The 2.18 results are more correct than the 2.16 results.

Thanks a lot for the quick reply. I'll adjust the test accordingly to fit the more correct results, Andreas.