samirelanduk / zincbinddb

The database and API backend for ZincBind - the database of zinc binding sites

ZincBindDB

About

ZincBindDB is a database of Zinc Binding Sites. It was made as part of my PhD, and is still maintained by me.

Together with ZincBindPredict (machine learning prediction of zinc binding using the data here) and ZincBindWeb (the website frontend) it comprises one of the three components of the overall ZincBind resource.

zincbinddb's Issues

Set Max Width for home page panels

On wide screens they are stretched out rather unnecessarily.

Image on home page has grey shadow at bottom

Only really noticeable on small screens

GUI selectors use chain of 'None' for waters

...which said GUI programs can't parse

Typo on main page

accross --> across

PDB superceding not accounted for

Unless the database is destroyed and rebuilt from scratch, there is currently no mechanism for removing PDB codes that have been superceded by another.

Species chart breaks if 'None' is on the x axis

Only really apparent while the database is building but as PDB grows this may change.

Add Second Shell information

This is vital for prediction

ZincBind should handle 'double zinc sites'

For example PDB 5EBB has six zinc sites, but in reality they seem to be three zinc sites with two zinc atoms in each.

Corners of 3D display are not rounded on certain versions of chromium

Big PDBs omitted

PDBs which only exist in mmCIF (etc) form are not currently obtainable or parsable, so some zinc sites may be missing.

Add tests for BLAST

Initially there weren't tests because it relies on a binary, but now that a wowrkaround for CD-Hit has been found, this should be fine too.

Add Jupyter notebooks for data visualisation

Erroneous Site residues

Currently main chain atoms are used to determine if a site is liganding - using only side chains might make more sensible sites.

Incorrect residues

In 3SEYE373, one residue is selected which is clearly not part of the site.

Move data folder on server

Currently uploading new code deletes all data, requiring data to be uploaded separately afterwards.

ZincSites with zero residues

Some zinc atoms just aren't close enough to other molecules to have a binding site allocated. Currently, it will still be saved, but as a ZincSite with zero residues.

This needs to be handled better, and maybe even have them ignored altogether.

NGL display doesn't handle insertion codes

1YEGL217 for example looks absurd.

Multiple Occupancy not handled in NGL

For example - 3SEYC378

Search by residue

It might be useful to search by the kind of residues involved - such as 'show me ones made only of cysteine' for example.