Delineating Protein Structural Features and Relationships Without PDB Homology Examples Using NCOR1

1
Delineating Protein Structural Features and
Relationships Without PDB HomologyExamples
Using NCOR1 and AdiponectinBy Chris Southan,
Bioinformatics GroupJune 2005
2
There Are Complete or Redundancy-Reduced Search
Spaces You Can Check

• Structure are ramping up but still 25 x more
  sequences in Uni90 than
• the 9432 90 identity PDB subset
• ( you can get this from OCA
  http//bip.weizmann.ac.il/oca-bin/ocamain a nice
  PDB front-end)
• UniProt Release 5.2 consists of 1,963,785 entries
  (UniProt/Swiss-Prot 184,304 entries and
  UniProt/TrEMBL 1,779,481 entries)
• UniRef100 Release 5.2 consists of 2,391,863
  entries
• UniRef90 Release 5.2 consists of 1,520,473
  entries
• UniRef50 Release 5.2 consists of 810,041 entries
• UniParc Release 5.2 consists of 5,299,541 entries
  
• You can search all these at
  http//www.expasy.org/tools/blast/ or
• http//www.ebi.ac.uk/fasta33/

3
First Stop SwissProt and InterPro
http//ch.expasy.org/sprot/ and
http//www.ebi.ac.uk/interpro/

• InterPro pre-cooks all family and domain
  relationships that can be recognised above a
  useful specificity threshold currently 92 of
  UniProt/Swiss-Prot and 78 of UniProt have a
  family and/or domain annotation including PDB
  homologues if they exist

HTRA4 protease
4
InterPro adiponectin/hemolysin relationships

• InterPro had captured this homology without the
  necessity to search the sequences
• Q86V24 ADR2_HUMAN Adiponectin receptor protein 2
• Q86WK9 Membrane progestin receptor alpha
• Q8XKC6 Clostridium hemolysin
• Q9VFG7 Drosophila membrane protein
• Q9U1V1 C.elegans membrane protein  
• All have IPR004254 HlyIII_related Pfam domain
  
• But you can get false positives like this in
  ADR2

5
Graphical Representations are Informative
http//www.ensembl.org/Homo_sapiens/protview?pepti
deENSP00000329071dbcore
http//www.cbs.dtu.dk/services/TMHMM/
http//www.ebi.ac.uk/interpro/ISpy?modesingleac
Q86WK9
6
Combined Representations are Even more Informative
7
Secondary Structure Prediction

• Now more powerful because of
• many orthologues for multiple alignments
• consensus combinations
• This is part of the structure prediction for
  Swiss-Prot entry NCOR1_HUMAN O75376 Nuclear
  receptor corepressor 1
• The website is http//npsa-pbil.ibcp.fr/cgi-bin/np
  sa_automat.pl?page/NPSA/npsa_seccons.html

8
NCOR1 - Combining Domain Mark-Up, Prediction and
Alignment
9
Good Prediction Tools at http//www.cbs.dtu.dk/ser
vices/but only 10 pops a day!