Interpin is an database that holds predictions for intrinsic transcription terminators called Hairpins in bacterial genomes. The database has predictions from 12745 bacterial genomes are placed. It can be found at Interpin db.
Here we provide the code used to make the predictions, along with this document that provides an easy installation guide for the same. The code can be used to make predictions for a single genome as well as run parallely on multiple genomes.
- Python3 (also install the following packages: matplotlib, biopython 1.78, multiprocessing, subprocess ). Anaconda / miniconda can be used for installation.
- Perl5
- Download and install edirect.sh Following command can be used:
sh -c "$(wget -q ftp://ftp.ncbi.nlm.nih.gov/entrez/entrezdirect/install-edirect.sh -O -)"After installation, you may add edirect commands to bashrc file (to access the program from anywhere). Following line should be added:
export PATH=${PATH}:$HOME/edirectThe same line can be used as a command to add edirect temporarily to current session only.
- Download and install bacbio package (for initial genome file processing). Following command may be used:
pip install bcbio-gff- Download and install curl. Use curllink for installation instructions on desired system.
Note: When installing these packages using pip, same version of pip should be used as python. Eg. pip3.6 should be used with python3.6 for all other package installations, pip3.7 with python 3.7 and so on.
Create a folder 'interpin'. Now create another folder 'program' inside it. All script codes must be placed inside this folder. Follow the steps below to make hairpin predictions using the interpin algorithm:
- Make a list of bacteria for which you want to make hairpin predictions. Place the file in interpin folder, outside prog folder. A sample for this input file is given in the 'sample' folder. Also find the format below:
file name: baclist.txt
file format:
bacteria1_name ncbiid
bacteria2_name ncbiidNote: Bacteria names and NCBI Id must be tab separated. Each line is record for one genome and must end by a new line character('\n'). Please remove any empty line after the last bacteria as an empty line will cause error when running the program.
Note: In ncbi id please give the complete ncbi id along with the version eg. NC_014614.1: adding .1 is essebntial to get the correct file for program input.
Note: Do not use any space or special case character in bacteria name. replace them by a '_' or '-'. eg: Acetobacter ascendens LMG 1590 to Acetobacter_ascendens_LMG_1590
- Now, run program inifiles_download.py. This will create a folder : interpin/genomes
cd program
python inifiles_download.py --n numhere 'num' is the number of cores available for processing. For serial processing, use 'n=1'.
Note: to see requirements of program and see short description of the program, use the below command.
python inifiles_download.py -h- Download Molquest for making operon prediction, which will be used for creating boundaries of transcription units. Download from here: Molquest
Note: The free trial for Molquest is currently only available for Windows and Mac users.
Afer downloading Molquest, create a new project. Then select the FgenesB annotator module from the left side panel, by double clicking it. Now input genome fasta file as input, which can be taken from the required bacterial folder inside genomes folder.
As output, two files are given. The results files or the folder can be found by clicking the file symbol that appears on main window after molquest run is complete. Take the 'results.txt' file and place this prediction in the same bacterial folder from where fasta file was takenThis is the only file required from molquest.
- Download Mfold package for RNA foldings. Details about the software and installation intruction can be found here: Mfold download and about. It is important to add mfold bin folder to path variables so that Mfold program can be called from our code. Add to path link can be used as reference help for this.
Also download the following package for Mfold.
sudo apt-get install texlive-font-utils- Now, all the input files are ready and the main interpin code can be run. Use the command below:
python interpin.py --n numhere 'num' is the number of cores available for processing. For serial processing, use 'n=1'.
Note: to see requirements of program and see short description of the program, use the below command.
python interpin.py -hThe final output is given in the form of a csv file, placed in the 'iden_hairpin' folder of each genome.
The sample files for each step are provided in the folder 'sample'. This would help understanding the format of input files if required.
As an example, I have run Intrepin codes on two bacterial genomes and placed them in the sample folder. The different folders formed, with the raw intermediate files are placed in these folders. Due to space restraints only 50 files from each strand showing fasta sequence and mfold output have been shown in folders 'fasta_op' and 'det_files'. Rest of the raw data is complete.
See below image for explanation of output:
The description columns in the output table is shown:
| Operon coding start | operon coding end | Hairpin start | Hairpin end | energy | Hairpin type | No. of constituent hairpin | strand |
|---|---|---|---|---|---|---|---|
| 124 | 1446 | 1476 | 1487 | 0.8 | single | 1 | forward |
| 1572 | 1498 | 1489 | 1415 | -5.7 | cluster | 3 | reverse |
The first row is a single hairpin at the end of operon [124, 1446] and is located at [1476, 1487] on the forward strand, with energy 0.8 Kcal/mol.
The second row is a cluster hairpin (with 3 constituent hairpins), at the end of operon [1572, 1498] and located at [1489, 1415] on the reverse strand, with energy -5.7 Kcal/mol.
- Can be run parallely on multiple genomes (number depends on cores available)
- Takes 10-15 hours for a giving predictions, with 90% time taken by Mfold to make folded structures.
- Cross platform
- No bias for AT/ GC rich genomes
- Predicts cluster as well as single hairpin. Cluster hairpin are novel type of hairpins given by the algorithm. To know more about the study, you can check out our publication here
If you use this code, please cite "Gupta, S., Pal, D. Clusters of hairpins induce intrinsic transcription termination in bacteria. Sci Rep 11, 16194 (2021). https://doi.org/10.1038/s41598-021-95435-3"
I'm a PhD student in the field of Computational biology at the Computational and Data sciences department, Indian Institute of Science. I have worked with protein, RNA and DNA sequences, structure and annotations, alignment, docking etc.
Currently I work with genomic data and find interesting patterns in them. I then try to find the underlying principles for those patterns.