Inputs and outputs
The FPS server takes as input a single protein sequence, GCG profile or BLAST checkpoint file, and returns a list of the families to which the query most likely belongs. This page provides details about the input format required by the server and the output produced by the server.
Input to Family Pairwise Search
To use FPS you must specify- a query, and
- a protein family library.
The Query
The query may be either a
sequence,
GCG profile, or a
BLAST checkpoint file.
Optionally, the sequence may begin with a single FASTA header line,
beginning with the ">" character. Here is an example query sequence:
The Protein Family Library
In addition, FPS requires you to select a library of
protein families to search. The FPS server currently supports two
protein libraries:
An example FPS output page is available here. The output consists of two sections. The first section lists the
families that closely match the query. The families are
sorted by E-value. An E-value is just the p-value of the
query-to-family match multiplied by the number of families in the
library. The E-value of a query-to-family match therefore reflects
the probability of finding such a match by chance in the given family
library. Here is an example output table: The table of matches lists each family that matches the query with
an E-value less than 10. Each row in the table gives the name of the
family, the total number of sequences in the family ("size"), the
effective family size ("eff size"), and the E-value of the
query-to-family match. The effective family size is an estimate of
the degree of similarity among the family members.
The second output section contains summaries of the pairwise
alignments generated by the Smith-Waterman algorithm. Each summary
shows pairwise alignments between the query and members of a
single family. These summaries are intended to provide an explanation
for the scores that FPS computes as well as insight into the
similarities between the query and members of each family.
No attempt has been made to produce true multiple alignments by
employing information about similarities among family members.
Therefore, the summaries are likely to contain more gaps than a true
multiple alignment. Here is part of an example pairwise alignment summary:
The alignment summary includes all of the information given
previously in the query-to-family table. In addition, the first block
of the summary lists the pairwise p-values between the query and each
member of the family. Each block of the summary begins with the query.
Subsequent lines in the block list the family member
sequence identifier, the starting position of this block within the
sequence, and the sequence alignment to the query. Amino acids in the
alignment are colored according to their biochemical
properties.
Note that the alignment summary only lists the most significant
alignments in the family. Thus, in the exmple alignment summary shown
above, the lipocalin family contains eleven members, but the summary
only contains the five most significant alignments. FPS was developed by Timothy Bailey at
the San Diego Supercomputer Center,
and by William
Grundy at the Department of
Computer Science and Engineering, University of California, Santa Cruz.
The FPS server is funded by the National Biomedical Computation
Resource. Please send comments and questions to Timothy
Bailey at tbailey@sdsc.edu.
Although the FPS algorithm can be used to classify DNA or protein
sequences, the FPS server currently accepts only proteins. The protein
sequence should be written in the standard IUPAC alphabet, including
the letters ACDEFGHIKLMNPQRSTVWY for amino acids and
BUXZ for ambiguities. The sequence may be in uppercase or
lowercase letters but may not include any punctuation other than
spaces or line breaks.
>ICYA_MANSE this is an example FASTA header line
GDIFYPGYCPDVKPVNDFDLSAFAGAWHEIAKLPLENENQGKCTIAEYKYDGKKASVYNS
FVSNGVKEYMEGDLEIAPDAKYTKQGKYVMTFKFGQRVVNLVPWVLATDYKNYAINYNCD
YHPDKKAHSIHAWILSKSKVLEGNTKEVVDNVLKTFSHLIDASKFISNDFSEAACQYSTT
YSLTGPDRH
Sequences are filtered for low-complexity regions using the
"seg" program by default. This can be disabled by un-checking this
checkbox. Low complexity regions commonly give spuriously high
pairwise alignment scores that reflect compositional bias rather than significant
position-by-position alignment. Filtering can elminate these potentially confounding
matches (e.g., hits against proline-rich regions).
See "Statistics of local complexity in amino acid sequences and sequence databases" by
Wootton, J. C. and S. Federhen (1993), Computers in Chemistry 17:149-163, for
more information on filtering and the "seg" program.
The pairwise search will use the scoring chosen from the list.
Alternatively, you may specify the name of a file containing a GCG
v1.0 or v2.0 peptide profile as the query.
The contents of the file will be uploaded and used to search the specified
protein library. Profiles in v2.0 format are converted to v1.0 format
automatically. Multiple profiles in v2.0 files are concatenated into a
single v1.0 profile by FPS.
Alternatively, you may specify the name of a file containing a BLAST
v 2.0 checkpoint file. The contents of the file will be uploaded, converted
to a GCG v1.0 profile, and used to search the specified protein library.
The conversion to a profile consists of dividing the amino acid
position-specific frequencies contained in the checkpoint file by
the amino acid background frequencies used by PSI-BLAST,
and taking the logarithm.
The amino acid frequencies used by PSI-BLAST are:
A C D E F G H I K L M N P Q R S T V W Y
0.078 0.019 0.054 0.063 0.039 0.074 0.022 0.052 0.057 0.090 0.022 0.045 0.052 0.043 0.051 0.071 0.059 0.064 0.013 0.032
The PROSITE and PFAM databases were both purged by removing sequences until
no pairs of remaining sequences in them
have BLAST similarity scores greater than 250 using BLOSUM 62 scoring.
This was done using the PURGE program described in:
Neuwald, Liu, & Lawrence (1995) "Gibbs motif sampling:
detection of bacterial outer membrane protein repeats"
Protein Science 4, 1618-1632.
Output from Family Pairwise Search
SCOP Superfamily size eff size E-value
Lipocalins 11 4.63 1.34e-12
PDZ domain 1 1.00 2.69e+00
Catalytic domain of malonyl-CoA ACP transacylase 1 1.00 3.90e+00
Protocatechuate-3,4-dioxygenase, alpha and beta chains 2 1.90 5.47e+00
A sigma70 subunit fragment from RNA polymerase 1 1.00 7.47e+00
RecA protein, C-terminal domain 1 1.00 7.53e+00
Acylphosphatase 1 1.00 8.57e+00
alignment of ICYA_MANSE vs. Lipocalins
family size: 11
effective family size: 4.63
p-value of match: 2.32e-15
E-value of match: 1.34e-12
ICYA_MANSE - 2 DIFYPGYCPDVKPVNDFDLSAFAGAWHEI---A--------KLPLENENQG--KCTIAEY
d1bbpa_ 3.84e-29 1 NVYHDGACPEVKPVDNFDWSNYHGKWWEV---A--------KYPNSVEKYG--KCGWAEY
d1hbp__ 1.19e-05 14 NFDKARFAGTWYAM---A--------KKDPEGLFLQ--DNIVAEF
d1epba_ 1.59e-04 1 VKDFDISKFLGFWYEI---AfaskmgtpGLAHKEEKMG---AMVVEL
d1beba_ 9.97e-04 7 DIQKVAGTWYSL---A----------------MA--ASDISLL
d1obpa_ 7.34e-03 8 NLSELSGPWRTVyigS--------TNPEKIQENGpfRTYFREL
ICYA_MANSE 49 KYDGK---------KASV--YNSFVSNGVKEYMEGD--LEI----------APD-AKYTK
d1bbpa_ 48 TPEGK---------SVKV--SNYHVIHGKEYFIEGT--AYP----------VGD-SKI--
d1hbp__ 46 SVDENghmsatakgRVRL--LNNW---DVCADMVGT--FTD----------TEDpAKF--
d1epba_ 42 K--------------------ENLLALTTTYYSEDHcvLEK----------VTA-TEGDG
d1beba_ 29 DAQSA---------PLRV--Y----VEELKPTPEGD--LEIllqkwengecAQK-KIIAE
d1obpa_ 43 VFDDE---------KGTVdfYFSVKRDGKWKNVH---------------------VKATK
AMINO ACIDS COLOR
AFILMPVW RED
CGHNQSTY LIME
DE BLUE
KR MAGENTA
