Hi,
I'm not involved in FF development, but I'd like to add my 2 cents
here. I do agree with Yong, and there's probably an easy and quick
solution for that. I posted before a suggestion of printing a "header"
in the mdout file that contains citation information. That's easy to
do with the program citation, but can be harder withthe FF case. A
workaround could be to add citation information in the prmtop file,
and this info be read and written to the mdout file as well. This way,
the user will always have this information at hand.
That would also solve another difficulty, the problem of remembering
which ff you used a long time after the simulation was performed,
since this info will always be available in the prmtop and mdout
files.
Gustavo.
On 1/29/07, Yong Duan <duan.ucdavis.edu> wrote:
>
> Following the line of discussion on citations, I think we need to look into
> the citation issues of the force fields too. In my opinion, ff citation is
> actually even messier. We need to come up with a "standard" set of citations
> and may recommend citations of multiple papers when a particular flavor of
> ff is used.
>
> For example, I'd think a sensible set of citations of ff99 and ff96 should
> also include the ff94 paper. Similarly, ff03 and ff02 citations should
> include the ff99 paper because most torsion parameters were actually
> "borrowed" from the ff99 paper. If we do not use this "multi citation" set,
> we give users the wrong impression that the improved ff is a "new" one
> (which is, of course, partially true). The multi-paper citation also
> encourages the users to understand the progression and the genealogy of the
> ff.
>
> If we establish reasonable rules, we attribute the credit appropriately and
> everybody will be happy. This can encourage further development and reduce
> the level of confusion. Regardless how careful or how deligent we are at the
> development stage, further improvements will likely be needed because, as we
> gather more data and experience, it will be inevitable to adjust a few
> parameters to improve some aspects of the force field. Keep in mind,
> computer power doubles every 1.5 years (about 100 times every 10 years). So,
> whatever the test we do now will be considered somewhat "too little" 10
> years later simply because more can be done a lot more easily. A result of
> this is that we may need to update some of the parameters. Hopefully, if the
> force field is designed well, we only need to change a few parameters.
>
> Please let me know your opinion, particularly those involved in ff
> development.
>
> Yong
>
>
Received on Wed Jan 31 2007 - 06:07:23 PST
