mmcif_pdbx.dic 5.283

[Diego Gallego]

Good morning!

Today I noticed that the mmCIF format changes with time following the "mmcif_pdbx.dic" dictionary (http://mmcif.wwpdb.org/dictionaries/mmcif_pdbx_v50.dic/Index/). As a result, new mmCIF files lack 5 columns in the coordinates section:

_atom_site.Cartn_x_esd

_atom_site.Cartn_y_esd

_atom_site.Cartn_z_esd

_atom_site.occupancy_esd

_atom_site.B_iso_or_equiv_esd

Therefore, the read.cif function is not working with updated cif files.

e.g.

read.cif("1ajr") Note: Accessing on-line CIF file trying URL 'http://www.rcsb.org/pdb/files/1ajr.cif' Content type 'text/plain;charset=UTF-8' length unknown .......... [...] .......... downloaded 720 KB

Error in .read_cif(file, maxlines = maxlines, multi = multi) : Could not convert string to int ("

Hope to be helpful!

Diego

[Barry Grant]

Thanks Diego - that is very helpful. We will likely post a patch release for this soon.

[Lars Skjærven]

Temporarily fixed with commit 4e0faf8. Note that this new version will only deal with the new format pdbx/mmcif files with 5 columns less than earlier format.

TODO: read columns based on definitions listed in the cif file.

[Diego Gallego]

Thanks Lars,

If it helps, I found out that cif files have a field called "_audit_conform.dict_version" that indicates the mmcif_pdbx.dic version. Moreover, version 4.072 was the last formated CIF files with 26 columns, and from 4.073 onwards they have 21 columns. With an if statement it should be rather straightforward to solve the matter for all cases. Cheers!

Diego

[Victoria Isabel Ruiz Serra]

Hi,

I've noticed that there is no problem reading any .cif file. However, I need to check the separated assemblies of each .cif file (available here ftp://ftp.pdbj.org/pub/pdb/data/biounit/mmCIF/all/) and I get a similar error:

Could not convert string to int ("?")

Since I've noticed there are 26 columns, I've tried to read the file with an older version of the bio3d package but didn't work. Any suggestions?.

Thanks!

Victoria.