prof (1)


NAME

prof - secondary structure and solvent accessibility predictor

SYNOPSIS

prof [INPUTFILE+] [OPTIONS]

DESCRIPTION

Secondary structure is predicted by a system of neural networks rating at an expected average accuracy > 72% for the three states helix, strand and loop (Rost & Sander, PNAS, 1993 , 90, 7558-7562; Rost & Sander, JMB, 1993 , 232, 584-599; and Rost & Sander, Proteins, 1994 , 19, 55-72; evaluation of accuracy). Evaluated on the same data set, PROFsec is rated at ten percentage points higher three-state accuracy than methods using only single sequence information, and at more than six percentage points higher than, e.g., a method using alignment information based on statistics (Levin, Pascarella, Argos & Garnier, Prot. Engng., 6, 849-54, 1993). PHDsec predictions have three main features:
1. improved accuracy through evolutionary information from multiple sequence alignments
2. improved beta-strand prediction through a balanced training procedure
3. more accurate prediction of secondary structure segments by using a multi-level system

Solvent accessibility is predicted by a neural network method rating at a correlation coefficient (correlation between experimentally observed and predicted relative solvent accessibility) of 0.54 cross-validated on a set of 238 globular proteins (Rost & Sander, Proteins, 1994, 20, 216-226; evaluation of accuracy). The output of the neural network codes for 10 states of relative accessibility. Expressed in units of the difference between prediction by homology modelling (best method) and prediction at random (worst method), PROFacc is some 26 percentage points superior to a comparable neural network using three output states (buried, intermediate, exposed) and using no information from multiple alignments.

Transmembrane helices in integral membrane proteins are predicted by a system of neural networks. The shortcoming of the network system is that often too long helices are predicted. These are cut by an empirical filter. The final prediction (Rost et al., Protein Science, 1995, 4, 521-533; evaluation of accuracy) has an expected per-residue accuracy of about 95%. The number of false positives, i.e., transmembrane helices predicted in globular proteins, is about 2%. The neural network prediction of transmembrane helices (PHDhtm) is refined by a dynamic programming-like algorithm. This method resulted in correct predictions of all transmembrane helices for 89% of the 131 proteins used in a cross-validation test; more than 98% of the transmembrane helices were correctly predicted. The output of this method is used to predict topology, i.e., the orientation of the N-term with respect to the membrane. The expected accuracy of the topology prediction is > 86%. Prediction accuracy is higher than average for eukaryotic proteins and lower than average for prokaryotes. PHDtopology is more accurate than all other methods tested on identical data sets.

If no output file option (such as --fileRdb or --fileOut) is given the RDB formatted output is written into ./INPUTFILENAME.prof where 'prof' replaces the extension of the input file. In lack of extension '.prof' is appended to the input file name.

Output format

The RDB format is self-annotating, see example outputs in /usr/share/profphd/prof/exa.

REFERENCES

Rost, B. and Sander, C. (1994a). Combining evolutionary information and neural networks to predict protein secondary structure. Proteins, 19(1), 55-72.
Rost, B. and Sander, C. (1994b). Conservation and prediction of solvent accessibility in protein families. Proteins, 20(3), 216-26.
Rost, B., Casadio, R., Fariselli, P., and Sander, C. (1995). Transmembrane helices predicted at 95% accuracy. Protein Sci, 4(3), 521-33.

OPTIONS

See each keyword for more help. Most of these are likely to be broken.
a
alternative connectivity patterns (default=3)
3
predict sec + acc + htm
acc
predict solvent accessibility, only
ali
add alignment to 'human-readable' PROF output file(s)
arch
system architecture (e.g.: SGI64|SGI5|SGI32|SUNMP|ALPHA)
ascii
write 'human-readable' PROF output file(s)
best
PROF with best accuracy and longest run-time
both
predict secondary structure and solvent accessibility
data
data=<all|brief|normal|detail> for HTML out: only those parts of predictions written
debug
keep most intermediate files, print debugging messages
dirWork
work directory, default: a temporary directory from File::Temp::tempdir. Must be fully qualified path.

Known to work.

doEval
DO evaluation for list (only for known structures and lists)
doFilterHssp
filter the input HSSP file (excluding some pairs)
doHtmfil
DO filter the membrane prediction (default)
doHtmisit
DO check strength of predicted membrane helix (default)
doHtmref
DO refine the membrane prediction (default)
doHtmtop
DO membrane helix topology (default)
dssp
convert PROF into DSSP format
expand
expand insertions when converting output to MSF format
fast
PROF with lowest accuracy and highest speed
fileCasp
name of PROF output in CASP format (file.caspProf)
fileDssp
name of PROF output in DSSP format (file.dsspProf)
fileHtml
name of PROF output in HTML format (file.htmlProf)
fileMsf
name of PROF output in MSF format (file.msfProf)
fileNotHtm
name of file flagging that no membrane helix was found
fileOut
name of PROF output in RDB format (file.rdbProf)

Known to work.

fileProf
name of PROF output in human readable format (file.prof)

Broken.

fileRdb
name of PROF output in RDB format (file.rdbProf)

Known to work.

fileSaf
name of PROF output in SAF format (file.safProf)
filter
filter the input HSSP file (excluding some pairs)
good
PROF with good accuracy and moderate speed
graph
add ASCII graph to 'human-readable' PROF output file(s)
htm
use: 'htm=<N|0.N>' gives minimal transmembrane helix detected default is 'htm=8' (resp. htm=0.8) smaller numbers more false positives and fewer false negatives!
html argument
'hmtl' or 'html=<all|body|head>' write HTML format of prediction 'html' will result in that the PROF output is converted to HTML 'html=body' restricts HTML file to the HTML_BODY tag part 'html=head' restricts HTML file to the HTML_HEADER tag part 'html=all' gives both HEADER and BODY
keepConv
keep the conversion of the input file to HSSP format
keepFilter argument
<*|doKeepFilter=1> keep the filtered HSSP file
keepHssp argument
<*|doKeepHssp=1> keep the intermediate HSSP file
keepNetDb argument
<*|doKeepNetDb=1> keep the intermediate DbNet file(s)
list argument
<*|isList=1> input file is list of files
msf
convert PROF into MSF format
nice
give 'nice-D' to set the nice value (priority) of the job
noProfHead
do NOT copy file with tables into local directory
noSearch
short for doSearchFile=0, i.e. no searching of DB files
noascii
surpress writing ASCII (i.e. human readable) result files
nohtml
surpress writing HTML result files
nonice
job will not be niced, i.e. not run with lower priority
notEval
DO NOT check accuracy even when known structures
notHtmfil
do NOT filter the membrane prediction
notHtmisit
do NOT check whether or not membrane helix strong enough
notHtmref
do NOT refine the membrane prediction
notHtmtop
do NOT membrane helix topology
nresPerLineAli
Number of characters used for MSF file. Default: 50.
numresMin
Minimal number of residues to run network, otherwise prd=symbolPrdShort. Default: 9.
optJury
Adds PHD to jury. Default: `normal,usePHD'.

Many other parameters change the default for this one as a side-effect, the list is not comprehensive:

phd, nophd, /^para(3|Both|Sec|Acc|Htm|CapH|CapE|CapHE)/, /^para?/, jct

para3
Parameter file for sec+acc+htm. Default: `<DIRPROF>/net/PROFboth_best.par'.
paraAcc
Parameter file for acc. Default: `<DIRPROF>/net/PROFacc_best.par'.
paraBoth
Parameter file for sec+acc. Default: `<DIRPROF>/net/PROFboth_best.par'.
paraSec
Parameter file for sec. Default: `<DIRPROF>/net/PROFsec_best.par'.
riSubAcc
Minimal reliability index (RI) for subset PROFacc. Default: 4.
riSubSec
Minimal reliability index (RI) for subset PROFsec. Default: 5.
riSubSym
Symbol for residues predicted with RI < riSubSec/Acc. Default: `.'.
s_k_i_p
problems, manual, hints, notation, txt, known, DONE, Date, date, aa, Lhssp, numaa, code
saf
convert PROF into SAF format
scrAddHelp
scrGoal
neural network switching
scrHelpTxt
Input file formats accepted: hssp,dssp,msf,saf,fastamul,pirmul,fasta,pir,gcg,swiss
scrIn
list_of_files (or single file) parameter_file
scrName
prof
scrNarg
2
sec
predict secondary structure, only
silent
no information written to screen - this is the default
skipMissing
do not abort if input file missing!
sourceFile
prof
test
is just a test (faster)
translate-jobid-in-param-values
String 'jobid' gets substituted with $par{jobid}
tst
quick run through program, low accuracy
user
user name
--version
Print version

AUTHOR

B. Rost, Sander C, Fariselli P, Casadio R, Liu J, Yachdav G, Kajan L.

EXAMPLES

Prediction from alignment in HSSP file for best results
 prof /usr/share/profphd/prof/exa/1ppt.hssp fileRdb=/tmp/1ppt.hssp.prof

Prediction from a single sequence
 prof /usr/share/profphd/prof/exa/1ppt.f fileRdb=/tmp/1ppt.f.rdbProf

phd.pl invocation
 /usr/share/profphd/prof/embl/phd.pl /usr/share/profphd/prof/exa/1ppt.hssp htm fileOutPhd=/tmp/query.phdPred  fileOutRdb=/tmp/query.phdRdb  fileNotHtm=/tmp/query.phdNotHtm

ENVIRONMENT

PROFPHDDIR
Override package prof package dir /usr/share/profphd.
RGUTILSDIR
Override location of librg-utils-perl /usr/share/librg-utils-perl.

FILES

*.rdbProf
default output file extension
/usr/share/profphd/prof
default data directory

BUGS

Please report bugs at <https://rostlab.org/bugzilla3/enter_bug.cgi?product=profphd>.
Prediction from HSSP file fails when residue lines with exclamation marks `!' are present:
Use 'optJury=normal' and 'both' like this:

 prof /tmp/1a3q.hssp fileRdb=/tmp/1a3q.hssp.profRdb optJury=normal both

SEE ALSO

Main website
<http://www.predictprotein.org/>
Documentation
<http://www.predictprotein.org/docs.php>
Community website
<http://groups.google.com/group/PredictProtein>
FTP
<ftp://rostlab.org/pub/cubic/downloads/prof>
Newsgroups
<http://groups.google.com/group/PredictProtein>