This repository contains the C++ source code for the LinearTurboFold project, an end-to-end linear-time algorithm for structural alignment and conserved structure prediction of RNA homologs, which is the first joint-fold-and-align algorithm to scale to full-length SARS-CoV-2 genomes without imposing any constraints on base-pairing distance.

LinearTurboFold: Fast Folding and Alignment for RNA Homologs with Applications to Coronavirus

Sizhen Li, He Zhang, Liang Zhang, Kaibo Liu, Boxiang Liu, David Mathews*, Liang Huang*

* corresponding authors

gcc 4.8.5 or above;
python2.7

Make

LinearTurboFold can be run with:

./linearturbofold -i input.fasta -o output_dir [OPTIONS]

The input file should be in the FASTA format. Please see input.fasta as an example.
Output a multiple sequence alignment and predicted secondary structures in the output directory.

--it The number of iterations (default 3).
--b1 The beam size for LinearAlignment (default 100, set 0 for infinite beam).
--b2 The beam size for LinearPartition (default 100, set 0 for infinite beam).
--pf Save partition functions for all the sequencs after the last iteration (default False).
--bpp Save base pair probabilities for all the sequencs after the last iteration (default False).
-v Print out alignment, folding and runtime information (default False).
--th Set ThreshKnot threshknot (default 0.3).
--tkit Set ThreshKnot iterations (default 1).
--tkhl Set ThreshKnot minimum helix length (default 3).

./linearturbofold -i input.fasta -o results/ --pf --bpp
100% [==================================================]
3 iterations Done!
Outputing partition functions to files ...
Outputing base pair probabilities to files ...
Outputing multiple sequence alignment to results/output.aln...
Outputing structures to files ...

We used the RNAStralign dataset with known alignments and structures to evaluate LinearTurboFold and benchmarks.

The 25 SARS-CoV-2 and SARS-related genomes analyzed in the paper are listed in samples25.fasta.
For further study by experts, we provide the whole multiple sequence alignment and predicted structures for all genomes from LinearTurboFold in sars-cov-2_and_sars-related_25_genomes_msa_structures.txt.
Each genome corresponds to three lines: sequence name, aligned sequence and aligned structure, respectively.