Jun 04, 2018

Public workspaceIdentification of proteins containing transmembrane domains using Phobius

DOI
dx.doi.org/10.17504/protocols.io.pdwdi7e
  • 1Beja Lab
José Flores-Uribe
Icon indicating open access to content
QR code linking to this content
DOI: dx.doi.org/10.17504/protocols.io.pdwdi7e
Protocol Citation: José Flores-Uribe 2018. Identification of proteins containing transmembrane domains using Phobius. protocols.io https://dx.doi.org/10.17504/protocols.io.pdwdi7e
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License,  which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Working
We use this protocol and it's working
Created: April 12, 2018
Last Modified: July 03, 2018
Protocol Integer ID: 11414
Keywords: transmembrane domains, phobius
Guidelines
This protocol was executed via ssh on a Ubuntu machine using the command line.
Before start
Checklist:
1. FASTA file containing the proteins to analyze.
2. FASTX toolkit installed
3. phobius installed
4. pandas installed in python
Gathering the sequence from the proteins to analyze
Gathering the sequence from the proteins to analyze
If you already possess a file containing the proteins to screen in FASTA format, skip this step.
Here we will retrieve all the proteins from the Organic Lake phycodnavirus 2 (OLPV2) from the NCBI website.
To download the proteins in the GenBank entry:
1. Click 'Send to:'
2. Select 'Coding Sequences'
3. Select 'FASTA Protein' as Format.
4. Create File
As shown below:


Then rename the file generated to something informative like:

OLPV2_prot.txt

Link to the GenBank entry for Organic Lake phycodnavirus 2:
https://www.ncbi.nlm.nih.gov/nuccore/HQ704803.1
Expected result
FASTA format file containing the proteins to analyze.
Reformatting the protein FASTA file.
Reformatting the protein FASTA file.
The FASTA file will be converted into a TSV file with two columns: 
1. The sequence header identifier
2. The amino acid sequence
The tool we will use is fasta_formatter from the FASTX toolkit.
Command
Convert a FASTA format file into a tab separated one. (Ubuntu 14.04.4 LTS)
fasta_formatter -t -i OLPV2_prot.txt -o OLPV2_prot.tsv
Expected result
A two columns file where the first column contains the sequences header and the second the sequence.
Software
FASTX Toolkit
NAME
Ubuntu 14.04.4 LTS
OS
Assaf Gordon
DEVELOPER
Obtaining the Phobius predictions
Obtaining the Phobius predictions
Here we will relly on Phobius combined with some command line tools (tail, tr, and awk) to filter the Phobius results and retain only those proteins with at least 1 transmembrane domain.
For an explanation of each part of the command read the steps below, otherwise skip to the next section.
Command
phobius.pl -short OLPV2_prot.txt | tail -n+2 | tr -s ' ' | tr ' ' '\t' | awk -F '\t' '$2 > </ProtocolCommand> 
<ProtocolResult	result=
Command
The -short option of phobius prints the results in a condensed table form with the following columns: seq_id, #TM, SP, description. 
phobius.pl -short OLPV2_prot.txt
Command
Ignores the first line of the output produced by phobius
tail -n+2
Command
By default the Phobius output separates the columns using whitespaces ' ' which is great for quick visual comparison of values but annoying for programmatically processing the information, here using tr we replace every multiple appearance of ' ' by a single one. E.g. '1 2' --> '1 2'
tr -s ' '
Command
Next we replace the whitespaces by tabs
tr ' ' '\t'
Command
Finally using awk the results are filtered to include only those with TM domains. The $2 > 0 is the parameter indicates awk to retain only lines where the second column ($2 in awk terminology), the one where phobius prints the number of TM, shows presence of TM in the sequence.  The output of awk is piped into the tab-separated file OLPV_prot_TM.tsv
awk -F '\t' '$2 > 0' > OLPV_prot_TM.tsv
Merging the phobius predictions to the FASTA sequences
Merging the phobius predictions to the FASTA sequences
To merge the tables of sequences and Phobius results we will use the following python script.
Copy paste the following into a file called:

merge_tables.py

Command
#!/usr/bin/env python
import pandas as pd
import sys
phobius_table = sys.argv[1]
proteins_table = sys.argv[2]
merged_tables_file = 'merged_tables.tsv'
phobius_df = pd.read_table(phobius_table, sep='\t', names=['SEQ_ID', 'TM', 'SP', 'PREDICTION'])
proteins_df = pd.read_table(proteins_table, sep='\t', names=['SEQ_HEADER', 'SEQ'])
proteins_df['SEQ_ID'] = proteins_df['SEQ_HEADER'].apply(lambda x: x.split(' ')[0])
proteins_df['DESCRIPTION'] = proteins_df['SEQ_HEADER'].apply(lambda x: ' '.join(x.split(' ')[1:]))
merged_df = phobius_df.merge(proteins_df, on='SEQ_ID', how='left')
merged_df = merged_df.loc[merged_df['TM'] >= 5]
merged_df.to_csv(merged_tables_file, sep='\t', columns=['SEQ_ID', 'DESCRIPTION', 'TM', 'SP', 'PREDICTION', 'SEQ'])
Command
The script takes two arguments: 1. The tab-separated file containing the phobius results. 2. The tab-separated file containing the amino acid sequences.
python merge_tables.py OLPV2_prot_TM.txt OLPV2_prot.tsv
Expected result
A tab separated file called merged_tables.tsv with six columns:
1. SEQ_ID: Identifier for each sequence.
2. DESCRIPTION: if the analyzed proteins came from GenBank this field contains the annotations.
3. TM: Number of transmembrane domains identified by Phobius.
4. SP: Presence of signal peptide in the protein.
5. PREDICTION: The segments of the protein corresponding to the different transmembrane domains.
6. SEQ: Amino acid sequence of the protein