Calculation of van der Waals molecular volume and surface

The results for van der Waals molecular volume and surface as calculated using Marvin plugs-in, are not in agreement with other published data, such as those from: M. Petitjean, J. Comput. Chem., 15 (No. 5) (1994) 507-523.  A few examples of the discrepancies, in particular for the van der Waals surface areas are considerable as shown in the Table below. 

				Petitjean	Petitjean	Marvin	Marvin
No	CAS	Formula	Name	Vol vdW	Surf vdW	Vol vdW	Surf vdW
1	374-25-4	C4H3BrCIF3	4-bromo-3-chloro-3,4,4-trifluorobut-1-ene	112.349	144.91	119.35	175.84
2	378-1 3-2	C4H4Br2CIF3	1,4-dibromo-2-chloro-1,1,2-trifluorobutane	137.164	172.664	145.22	225.32
3	440-05-1	C11H15FO7	4,5-bis(acetyloxy)-6-fluorooxan-3-yl acetate	198.563	250.17	234.77	390.56
4	463-51 -4	C2H2O	ketene	38.552	58.486	40.77	65.74
5	504-64-3	C3O2	carbon suboxide	54.158	77.828	53.98	80.23
6	511-98-8	C27H45NO2	tomatidine	350.936	369.607	434.53	728.7
7	525-06-4	C14H10O	diphenylethenone	173.769	203.176	182.95	279.34
8	538-75-0	C13H22N2	N-(N-cyclohexylcarboximidoyl)cyclohexanamine	189.063	222.429	221.21	375.09
9	550-33-4	C10H12N4O4	nebularine	181.387	217.081	205.11	317.72
10	572-09-8	C14H19BrO9	acetobromo-α-D-glucose	270.156	330.612	310.36	505.53

Is there a specific explanation for such large discrepancies?  Some differences can be expected if the van der Waals atomic radii are slightly different in the two methods, but values larger by 60-70% than Petitjean reported numbers are not explaned by this.

 

Best regards,  Serban   

 

Dear Serban,

 

Could you send the 3D structures and the van der Waals atomic radii  used in the reference?

 

Regards,

Gabor

Hi,

 

Is there a specific explanation for such large discrepancies? 

Marvin's  surface area calculator is parametrized specifically for the "H" atoms.It is recomended for peptide and protein structures.   

From your table it is apparent that  the more  H atoms arre in the molecules the largest the difference between Petitjean's and Marvin's surface area. 

So , I collected   21 molecules from Petitjean's paper and a  correction function created from them in order to take into account the effect of the H count on the Van der Waals surface area.

Marvin's results are close to Petitjan's calculated results now. See in the attached file.

Unfortunately I was not able to find the large part of  the structures because their CAS Registry number were not available neither in the "Google search" or in the free database provided by Marvin's molconverter program .  Unused structures are denoted with a question mark in the attached file. 

My conclusion is that we need to implement an alternative method for surface area calculation of small molecules. We work on this now.

Thank you for reporting your results!

 Jozsi