Documentation top | FUGUE Home ]


How to interpret the output page

Introduction

Please first read pages 14-16 of the book chapter (available as PDF) for basic information. An example output file can be found here.

Examine your sequence and its sequence homologues

At the bottom of the header section, there are three lines:
The sequence(s) you submitted is HERE (in original format).
The sequence(s) actually used by FUGUE is HERE (in PIR format).
Download all the results in compressed format HERE. new!

If you click the first link, you should see the amino acid sequence that you submitted. Make sure this is what you wanted.

The second link shows a multiple sequence alignment of the homologous sequences collected by PSI-BLAST. PSI-BLAST is run against the NCBI nr database for up to five iterations with the default inclusion cut-off. These are the sequences that appear in the query-structure alignments of the 'aa' type (see below). To examine the individual homologous sequences, see How to examine homologous sequences below.

By clicking the third link, you can download all the results (including alignment files and PDB models) in a compressed archive. This file is generated by the UNIX tools 'tar' and 'gzip'. You will have to save it on disk first and extract the archive (using an appropriate tool).

Note for the command-line version: If you run the command-line program run_fugue with the default parameters, you will find the output file fugue.html. This file looks similar to the output web page but does not include the three lines described above. To obtain the information about the homologous sequences, look at the file name.inp, if the query sequence is saved in the file name.seq and the program was run as run_fugue -seq name.seq.

How to examine the query-structure alignments

The View Alignments section gives alignments between the query sequence and each of the top 10 hits in three formats: HTML for online browsing, postscript (PS) for printing and pure text format for use with other programs. The HTML and PS files are produced by the program JOY to annotate the alignments with structural environments. Here is key to JOY format and more information about the program can be found at http://mizuguchilab.org/joy/. See also Help on HOMSTRAD. For each hit, four types of alignments are available ('aa', 'ma', 'mh' and 'hh'). These are explained at the bottom of the output page.

How to examine rough models

The model column in the alignment section provides links to rough models for the query sequence. These models are built in the following way:
  1. For each HOMSTRAD entry hit, select the structure that has the highest PID to the query sequence (same as the "hh" type alignment).
  2. According to the FUGUE alignment (of the "hh" type), copy the main chain coordinates of aligned residues from the selected structure.
  3. Output the model. Any residue in the query, which is not aligned to a residue in the structure, is omitted from the model.
Each model simply gives an overview of how good the alignment coverage is. There is no sidechain information and no attempt is made to fill the gaps. It is often informative to compare the model with the original template structure (in HOMSTRAD) and see whether the alignment coverage is reasonable. For instance, if the missing parts cause extensive exposure of hydrophobic core in the template structure, the alignment or the homology detection may have serious problems. On the other hand, it may not be too bad if a relatively independent module is missing (e.g. one domain is missing from a two-domain structure).

How to examine homologous sequences

The alignments of the 'aa' type include all the homologous sequences of the query collected by PSI-BLAST. The sequence names can be cryptic but you can interpret them in the following way. First, ignore the first part of the sequence name (1_, 2_, etc). These sequential numbers have been added by FUGUE to distinguish different sequences. (Without this, programs such as CLUSTALX may not read in an alignment file.) The numbers after 'gi|' correspond to the NCBI gi numbers but they are TRUNCATED. To retrieve the correct gi numbers, go back to the link to the PSI-BLAST alignment explained above. In this PIR file, all the homologous sequences with the full gi numbers are shown. The order of the sequences are identical to that of the 'aa' alignment files. If you use a UNIX machine, you can save the PSI-BLAST alignment (for example, as T0137.inp), then type
fgrep '>' T0137.inp
This will give you a list of homologous sequences. Once you have obtained the correct gi numbers, you can retrieve more detailed information on each of the proteins by going to http://www.ncbi.nlm.nih.gov and search protein for a particular gi number.

On the alignment algorithms

FUGUE performs the z-score calculation and the alignment generation independently. The last two columns of the rank section indicate the final choices of the alignment algorithms: the one for the z-score calculation and the one for the alignment, respectively (shown in red in the example below).
Profile Hit      PLEN   RAWS   RVN   ZSCORE   ZORI   AL

hs1lr0a 123 327 418 36.49 37.39 03 CERTAIN Alignment
Possible choices are:
0 Global -- Modified Needleman & Wunsch
2 GloLocSeq -- Global algorithm but overhang in the sequence NOT penalized
3 GloLocPrf -- Global algorithm but overhang in the profile NOT penalized
For more information, see Automatic selection of alignment algorithms in Notes for the use of command-line version.
Last update: 15 Apr 2015

Valid XHTML 1.0!