Sequence validation

We check all sequences before running a search. In order to avoid problems with the validation of your sequence, you should use only plain, unformatted text. Here are some of the validation checks that we apply to sequences:

• sequence length must be less than 10,000 residues
• the sequence must be a protein sequence; nucleotide sequences will not be accepted
• only residue symbols allowed in the sequence; sequences containing other characters, such as "#" or "*", will not be accepted
• FASTA-header lines are accepted but will be removed

If you have problems getting your sequence to upload, please check that it passes all of these tests. Note that although we do allow FASTA-style header lines on a sequence, some characters in header lines can still cause the sequence to be rejected. If in doubt, please remove header lines before pasting in your sequence.

You can see an example of a sequence that will successfully pass all of the validation tests by clicking the Example button below the search form.

Search options

You can choose to search using either or both of the HMMs that we produce for each family. Choose Global to search against the HMM that is optimised to look for full length families, known as the "ls" model, or Local to search against the HMM that is optimised to find fragments of Pfam families in a sequence, the "fs" model.

Rather than viewing hits from just one of the two models, you can also choose Global & local (separately) to see the results from both "ls" and "fs" models, or Global & local (merged) to have the results from the two models merged.

When we build a Pfam family, we remove overlapping hits on the same sequence in one of three ways: by taking global hits first, taking local hits first, or by taking the highest scoring overlapping hit. The rule that we use is recorded for each family and if you choose Global & local (merged), we will use the same rule to resolve overlaps between hits in your search.

The default threshold for the HMM search is an E-value of 1.0, but you can also use the gathering threshold for each HMM, or you can specify your own E-value setting. Note that the E-value that you give must be positive.

By default the search will only look for Pfam-A families on your sequence but, by checking the box below, you can also search for Pfam-B hits.

---

You can enter either an accession or an identifier for a Pfam family. The search is case insensitive.

We calculate functional similarity using the method described here. We currently apply an arbitrary cut-off of 0.75 to the similarity score, to avoid generating excessive volumes of results.

File contents

We accept only protein sequences and your uploaded file must conform to a fairly strict interpretation of the FASTA file format. We apply the following checks to the format of uploaded sequence files. Files that do not conform to the following rules will be rejected by the server

File contents

Files must contain only header lines and sequence lines. Header lines, which begin with ">", can be used to describe the sequence that follows. There is no fixed format for header lines but we restrict the characters that are allowed. If your header lines contain any of the following characters, your file will be rejected:

; \ ! *

Note that we explicitly include the semi-colon (;) in the list of forbidden characters, although this may be used to denote comments in some versions of the FASTA-format. Please do not use comments in the FASTA files that you upload here.

Header rows

The FASTA-format specification recommends that header lines are kept shorter than 80 characters. Batch searches are run using our pfam_scan.pl script, which uses programs from the HMMER suite. Because of the way that HMMER handles header lines in FASTA files, only the first 60 characters are actually used. Please make sure that your header lines are 60 characters or less in length.

In order to avoid generating strange or ambiguous search results, we require that header lines in your file are unique, at least within the first 60 characters. Uploaded files which contain duplicate header lines will be rejected.

Sequence symbols

Your sequence should be a valid protein sequence. As such, the sequence line should contain only amino-acid symbols, i.e. capital letters excluding "J". In the context of a Pfam search, gaps and translation stops have little meaning and should not normally be used, but we accept "-" or "*" to denote gaps and translation stops respectively.

Service limits

Searches run on a "compute farm" with a limited number of "slots". Each search takes one slot and once all slots are in use, new jobs wait in a queue for the next slot to become free. In order to prevent large jobs occupying slots for very long periods, which can impact the availability of the system for other users, we place a number of restrictions on the size of job that we will accept.

File size

Files must have fewer than 10,000 lines and fewer than 1000 sequences.

Sequence length

Each sequence must be no longer than 2000 amino-acids.

Sequence variation

We use heuristics to check that a sequence has a reasonable level of variation, in order to prevent large strings of identical sequence or a large number of duplicate residues being searched. If you find that you cannot submit a valid sequence because of this restriction, please let us know.

E-value limit

If you specify an E-value cut-off for your search, that E-value must be a positive number.

---

The search currently covers the following sections of the database:

• text fields within Pfam entries, such as description and comments
• sequence description and species fields
• the HEADER and TITLE records from PDB files
• Gene ontology (GO) IDs and terms
• InterPro abstracts

You can perform the same search from anywhere within the Pfam site, using the keyword search box at the top right-hand side of every page.

---

You can specify one or more domains that must appear in an architecture and one or more domains that must not appear in the result architectures:

1. choose a group of families or the first letter of the family ID from the list below
2. in the left-most list, select the ID of a domain
3. choose Includes or Does NOT include to add the selected domain to the appropriate list
4. continue until the Includes and Does NOT include lists show all of the domains that you need
5. press Submit query to search for architectures matching your specification

This tool allows you to search DNA sequences of upto 80kb against Pfam HMMs, using the Wise2 package. You can see more information about Wise2 at the homepage.

Because these searches can be time consuming, we run the searches offline and mail the results to you. Please ensure that the email address that you enter is correct, as no checking is performed before the search is run.

---

Species names

In order to find species names in our database, you must give the full, unabbreviated name in your query. For example, if you want to find families in Caenorhabditis elegans, you must spell out "Caenorhabditis"; "C. elegans" does not exist in the species database and will not return any results. The capitalisation of your query does not affect the results: "Caenorhabditis" is the same as "caenorhabditis".

Term suggestions

Because species names can be long, complex and easily misspelt, the form will offer suggestions for possible completions of terms that you enter. For example, as you start typing c-a-e-n, a drop-down panel will appear, showing the possible completions:

As you enter more letters, the suggestions will be refined accordingly:

Note that only the last term in the field is used to offer completions, so if you have already entered "caenorhabditis elegans", going back and shortening the first word to give "caen elegans" will not give any suggestions.

Boolean queries

Boolean queries can contain logical operators, such as AND, NOT and OR, combined with braces ("(" and ")") to form a description of a set of families with a particular species distribution.

For example, the following query:

Caenorhabditis elegans AND NOT Homo sapiens

will retrieve all families which are found in C. elegans but not in human, whilst

Caenorhabditis elegans AND Caenorhabditis briggsae AND NOT Homo sapiens

will retrieve all families which are found in both C. elegans and C. briggsae but not in human, and

( Caenorhabditis elegans OR Caenorhabditis briggsae ) AND NOT Homo sapiens

will retrieve families that are found in either C. elegans or C. briggsae but not in human.

Families unique to a single species

By checking the box marked "Find domains unique to query term" you can limit the query to finding those domains which are found only in a single species or taxonomic level. For example, checking the box and entering "Caenorhabditis elegans" will return a list of Pfam-A domains that are present only in C. elegans, whilst "metazoa" will find families that exist in metazoans.

Note that you can only enter a single species term when looking for unique sequences. You will see an error message if the search field contains more than one species.

---