NEWS

pqsfinder 2.2

Default minimal PQS score was decreased from 52 to 47. The score 47 shows the best balanced accuracy on new G4 sequencing data provided by Marsico et al. 2019.

Fixed bug allowing unlimited length of third loop leading to invalid memory access and random scores.

The qsfinder algorithm is ~100 times faster on real-world sequences than before. The main idea of the optimization is to stop the exhaustive search when we are sure there can't be found better PQS. The optimization is applied only in non-overlapping mode.
New boolean argument called "deep". Once set to TRUE, pqsfinder will disable optimizations and run full exhaustive search.

Default minimal PQS score was increased from 26 to 52. The score 52 shows the best balanced accuracy on G4 sequencing data provided by Chambers et al. 2015.

Fixed resolution of overlapping sequences. When multiple same-scoring PQS overlap, the shortest one is chosen.
Fixed incorrect PQS metadata. When PQS was found on antisense strand, corresponding run lengths and loop lengths were confused.

Fixed uninitialised variable causing slightly stochastic behaviour.
Fixed bug blocking alternative PQS with zero-length loop to be found.
Fixed inefficient looping through long non-guanine sequence regions.
Updated default scoring parameters to slightly better-performing combination.
Removed bug-prone special support for edge mismatches.
Fixed integer division to floating point division in the computation of loop length mean.

Added estimated time to complete (ETTC) to the progress line.
New default values for scoring parameters as a result from training on G4-seq experimental data and co-testing on a set of known quadruplexes from literature.
New PQS metadata reported: number of tetrads, bulges, mismatches and loop lengths. To get this data, use elementMetadata accessor function.
New algorithm option: reporting of all overlapping PQS.
Score distribution is reported. For each sequence position you get maximal score of PQS that was found overlapping the position. For details, see scoreDistribution function.
Novel scoring parameter: exponent of loop length mean in the scoring equation to express non-linear dependency of the PQS propensity to the loop lengths.

Minimal loop length is set back to 0 by default, but only one zero-length loop is allowed by the scoring system.
Fixed unintentional cast of loop length mean factor and loop standard deviation factors from float to integer.

Minimal loop length is now set to 1 by default.
Removed linking dependency on the flowCore package.
Fixed bug responsible for a segmentation fault. Great credit belongs to Diego Andino for sending a bug report.

Novel algorithm for identification of potential intramolecular G-quadruplex (G4) patterns in DNA sequence.
Supports multiple defects in G-runs like bulges or mismatches.
Provides the most accurate results currently available.
Highly customizable to detect even novel G4 types that might be discovered in the future.