hey im no expert in this topic but I found three pdfs which u might find useful. let me know if ur problem was solved by this. 

http://digitalcommons.uri.edu/cgi/viewcontent.cgi?article=1005&context=che_facpubs

http://thet.uni-paderborn.de/fileadmin/thet/publikationen/pdf/Trans_aqueous_alcohol.pdf

http://people.virginia.edu/~lz2n/mse627/notes/Diffusion.pdf

As an exercise for your project, you may look at my calculation of the diffusion coefficient in my recent post "Exact non-equilibrium statistical mechanics in one dimension", available s a download from Research Gate. Good luck.

If I got it right, you are doing this GROMACS? in GROMACS all the analysis of this type are atom based and then for each molecule specified the center of mass for those atom belonging to that molecule are used. If you choose all (e.g. water) molecules from one type, the results is not different from selecting each molecule separately (doing analysis separate;y as well) and then averaging over them. Just note that for averaging over atoms in one molecule, GROMACS weights the values by atomic mass (and you can change that if you want).   

thank you for your answer, and sorry for my late reply. I was away for a while....

Dear Abolfazl,

yes, I underestand. but I still have the question that if I use  the following :

g_velacc -acflen 1001 -nonormalize -mol -n index.ndx -s topol.tpr -o vac.xvg

what should be the index file? just all atom numbers or molequle numbers?( which one of the method that you mentioned is used here?)

when I use all atom numbers, apparently it split them to correct molequle number. But still the answer is not good! It might be due to other things, but first i should now which index file should I use.

Best regards

Negar