Casual reminder that what many consider 'information' from #crystallography is actually a model. That is why structure factors must *always* be published, so that reviewers and anyone can see what lies behind the model. Without them, crystal structures are nothing more than 'pictures' in 3D. - ThreadSky

stecanossa.bsky.social • 93 days ago

Casual reminder that what many consider 'information' from #crystallography is actually a model. That is why structure factors must *always* be published, so that reviewers and anyone can see what lies behind the model. Without them, crystal structures are nothing more than 'pictures' in 3D.

Comments

stecanossa.bsky.social•93 days ago

(and you'll also make @ccdc.cam.ac.uk happy 😊)

stecanossa.bsky.social•93 days ago

This implies that the cif (crystallographic information file) must contain structure factors, otherwise is not a cif, but a cmf 🙃

andypurk.bsky.social•93 days ago

As a mostly macromolecular crystallographer, one of the minor annoyances is that the model and SF data remain in separate .cif files in this field, for historical reasons.

stecanossa.bsky.social•93 days ago

Uh, surprising... macromolecular community is normally the first to innovate, that's often my impression. You sure it's only historical? if it's linked to file sizes and nonsense like that (which probably once were all but nonsense, with lower memory availability etc.) this should be brought up

andypurk.bsky.social•93 days ago

I think it is mostly legacy software (and users) which needs them separate. The sf file is 5-10 times bigger than the model data but still in the MBs range. So course, many of the end users won't be interested in that part. The PDB has only insisted on mmcif for model upload fairly recently.

stecanossa.bsky.social•93 days ago

apparently MB is not negligible yet... as soon as it will, that should be changed. It's like ADP refinement: not many small molecule people need it for determining structures of drugs etc., but the laziness barrier for it is a click on a button, so everybody does it. (meaningfully or not, true)

andypurk.bsky.social•93 days ago

Compared to the raw synchrotron data (in the 10s of GB per dataset), MBs is nothing. Of course storage at the PDB is another matter.
One of the joys of more automated software, select an extra option for a minor improvement in stats.

stecanossa.bsky.social•93 days ago

I provide more details on how to make those representations of electron density from structure factors (which I highly encourage!) in the SI of this paper: https://pubs.acs.org/doi/10.1021/jacs.2c13731

Comments

Posting Rules

Reply