[gmx-users] index file for g_velacc to calculate diffusion coefficient in the bulk

[Abbasi, N.]

Dear all,
I want to calculate the diffusion coefficient in the bulk based on Green-Kubo relation. Now I want to use :
g_velacc -acflen 1001 -nonormalize -mol -n   index.ndx -s topol.tpr  -o vac.xvg
I have got  different  information about  index.ndx file . should  it contain molecule numbers? Or number of atoms which form a molecule(I used this one, answer was not good!)? How can I make an index file containing the molecule numbers? What should I do  if I want to choose center of mass of each molecule to get the best answer?
Thanks in advance for your guidance
Negar