Prediction of surface-exposed proteins in bacterial cells
Downloadinmembrane is a program to predict if a set of proteins are likely to be surface-exposed in a bacterial cell based on their predicted subcellular localization and membrane topology. It is designed to be relatively easy to install, with options for running the analysis using web services and/or locally installed sequence analysis packages.
inmembrane runs on Linux, OS X and other Unix-based systems. It has been tested on Ubuntu Linux (11.04+) and Mac OS X (10.6+). It hasn't been tested on Windows, and is unlikely to run without changes.
$ sudo pip install inmembrane
$ inmembrane_scan --test
$ inmembrane_scan my_gram_pos_proteome.fasta
(Some tests may fail if you don't have various external binaries installed, but don't despair - the only external program required by is HMMER v3, everything else uses a web service in the default configuration)
We envisage inmembrane being of use in three ways:
Quirk and dirty annotation of a proteome in order to cross-check your experiment (eg, gaining an overview of hits from a gram-positive bacterial cell-surface shaving experiment).
You want to tweak your own rule-based method for subcellular localization assignment from sequence. Check out our parameterization guide.
You want to make your own protein sequence analysis work-flow, and would like to use inmembrane as a useful Python starting point. Have a look at our programming model and API guide.
The cell-surface exposure algorithm used by the inmembrane 'gram-positive' protocol is based on the method described in (Barinov et al. 2009. Proteomics 9:61-73). The workflow involves annotating the provided protein sequences using a number of external programs and/or web services providing:
inmembrane simplifies collating the output and presenting an overview which allows biologists to immediately gain insight into the likely localization and surface exposure of their bacterial proteome of interest.
Installing and running inmembrane requires:
On Debian/Ubuntu systems, you can install the minimal dependencies using:
$ sudo apt-get install python-pip hmmer
Once these are installed on your system, inmembrane can be installed with:
$ sudo pip install inmembrane
Let's say you have all the protein sequences you want to analyze in a FASTA text file called my_proteome.fasta.
Inmembrane is run from the command line like:
$ inmembrane_scan my_proteome.fasta
Assuming the web services inmembrane relies on are available, the analysis should complete within seconds or minutes.
A summary is printed to the terminal, and the results will also be written to my_proteome.csv, which can be opened by a spreadsheet application such as Microsoft Excel. Citations for all external programs and services are written to my_proteome/citations.txt - this is very handy and important when it comes to writing up a publication using the results of inmembrane and the software it orchestrates.
Pro-tip: The directory created by inmembrane (eg named my_proteome), as well as containing the citations.txt list, also contains cached output files from the individual external analyses, the original configuration file and the input FASTA file. If you run inmembrane a second time on the same FASTA file, the results will be read from this directory rather than re-running each external analysis program. To re-run the analysis from scratch (with say, different parameters), you must remove (or rename) this directory. To re-run just part of the analysis you can remove individual files from this directory.
It should look something like this:
SPy_0008 CYTOPLASM(non-PSE) . SPy_0008 from AE004092
SPy_0010 PSE-Membrane tmhmm(1) SPy_0010 from AE004092
SPy_0012 PSE-Cellwall hmm(GW2|GW3|GW1);signalp SPy_0012 from AE004092
SPy_0016 MEMBRANE(non-PSE) tmhmm(12) SPy_0016 from AE004092
SPy_0019 SECRETED signalp SPy_0019 from AE004092
There are 4 columns to the result:
tmhmm, hmm, and signalp