tdb webserver trucating my protein

[kad-ecoli]

I have tried to us the tdb webserver for one protein, which is 258 amino acids long, The result is at
http://bioinf.cs.ucl.ac.uk/bio_serf/getresultattached/158534327
which is 9 residues long.
Also, it is inconvenient to use a webserver to construct tdb file for hundreds of proteins on a weekly basis. I wondered whether there is any plan to release the programs used for TDB file generation, specifically that used for assigning relative solvent accessibility.

[DanBuchan]

Hi,
There are no plans to release the TDB generation script. I'll track down the bug on the webserver though.

In the meantime you can download the current TDB files. TDB files represent sequence clusters so you could blast your proteins aginst the TDB sequences to work out which cluster/fold your proteins belong to.

[kad-ecoli]

The fact that TDB files represent sequence clusters is another problem for us. It seems like the PDB database is clustered with very stringent sequence identity (30% or 40%) when the TDB library is created, compared to HHsearch library or I-TASSER library which are clustered by 70%. This means the TDB library does not well cover the full sequence space of PDB database (although it does cover the full structure space). Many times if I blast my protein against TDB, I cannot find anything. But if I blast against HHsearch's pdb70 I do find homolog.

[DanBuchan]

Sure. Genthreader isn't really a homologue detection tool. It is a fold recognition tool and the library is clustered for that purpose.

[DanBuchan]

Hi, with regards the TDB generator truncating your protein. Hopefully in the next month I'll have a fixed TDB generator ready