Pfam DATABASE

Pfam database uses Hidden Markov Models (HMM's)
HMM's are statistically based mathematical treatments, consisting of linear chains of match, delete or insert states that attemt to encode the sequence conservation with in aligned families.
Pfam database has a collection of HMM's for a range of protein domains which is maitained at the Sanger center.
Pfam database is based on 2 distinct classes of alignments:
1) Hand edited sedd alignments : Tese are considered to be accurate and are used in Pfam-A.
2) Alignments derived by automatic clustering of SWISS-PROT: tHESE ARE LESS RELIABLE.

The high quality seed alignments are used to build HMM's to which sequences are automatically aligned to generate final full alignment.
If the alignments donot produce diagonostically sound HMM's, the seed is improved and gahtering process is iterated untill a good result is achieved.
The collection of seed and full alignments, the HMM's, database and literature cross references constitute the Pfam-A
All sequence domains that are not included in Pfam-A are automatically clustered and deposited in Pfam-B.

STRUCTURE OF Pfam-A Entries

AC PF00001
ID 7tm
DE 7 transmembrane receptor (rhodopsin Family)
AU ..
AL ..
AM ..
SE ..
DR ..
DR ..
GA ..
DR ..


The format is compatible with PROSITE.

A brief description about the Lines is given below:

ID Line: A single Keyword is used
AC Line: Has the Accession number. The form of Accession number is PF00000
DE Line: Provides description of the family and the title.
AU Line: Indicates the author of the entry.
AL Line: Methods used to create seed alignments are provided in the AL Line.
AM Line: Methods used to create full alignment
SE Line: Information about the Source database
DR Lines : contain the Database cross references.