Purpose

Hairpin is a program which predicts RNA genes in a genome ab-intio. It is based on the assumption that regions in the genome having more possible pairings than expected by chance, will possible code for structured RNA genes. The resulting *.bed-file can be uploaded as a track to a genome browser based upon the same genome assembly.

Requirements

This is intended to work for Python 3.5.0 or later. The required packages with their versions are listed in requirements.txt, run pip3 install -r requirements.txt in order to install them.

Usage

The program is run from the command line using the syntax (example):

python3 hairpin.py --n_randomizations 50 --min_window 4 --max_window 20 \
--min_length 20 --max_length 100 --min_motif_length 4 --chrom 'chr' \
--chromStart 1000 --q_crit 0.05 --res_prefix 'my_results' e_coli_genome.fasta

The arguments are described upon calling:

python3 hairpin.py --help

Also, you can use the functions and classes provided by the software as part of a custom script as done in the examples. Note though that the internal API is poorly documented, so issues and pull requests regarding its documentation are welcome.

More about the arguments

Except the input file, all options have default values. The input must be in fasta-format, where the sequence shows the coding DNA strand on a chromosome. An input file with multiple records is allowed, but only the first record will be analyzed. Nucleotide ambiguity codes are accepted, but ignored in the analyses. In order to provide specifications to the genome browser, the name of the chromosome or scaffold must be specified much as --chrom chr10. The default argument however, is OK for procaryotes only having a single genome. In case the record is not the entire chromosome, consider the following cases:

• The record does cover the start of the chromosome: Specify the start position (zero-indexed) as (for instance) --chromStart 999
• The record does not cover the end of the chromosome: No action required

Note on performance

This program uses parallelization, meaning it might take up huge amounts of the computer's working capacity. For large input files or many rounds of randomizations, running on a computer with many cores is recommended.

Output

The program outputs three files:

• (res_prefix)track.bed: The genomic track information, can be uploaded to a genomic track browser
• (res_prefix)randomized_result.txt: Tab delimited text file specifying the results when constructing randomized sequences
• (res_prefix)plot.pdf: Plot showing the number of matched bases for randomizations and reported significant intervals. The blue line correspond to the mean number of matched bases for the randomizations, while the error bar denotes its standard deviations. Each red dot corresponds to a significant interval found.

Disclaimer

Do not expect the program to yield any meaningful biological results. As the referenced paper states, the biological signal is typically too small to be separated from random noise. You are welcome to prove the author wrong though.

References

Hairpin: a tool for predicting structural non-coding RNA genes ab initio Jakob Peder Pettersen bioRxiv 640862; doi: https://doi.org/10.1101/640862