Nucleosome positioning prediction

Below is an annotated list of available software to predict nucleosome positioning from DNA sequence. This list  is being constantly updated, comments are very welcome. The database of experimental nucleosome positioning in different cell types is moved to a separate page. For protein-DNA interaction of non-histone proteins and TFs, see the section on TF-DNA binding. Also, have a look at the epigenetic modifications section of the site.

*How to cite: Teif V.B. (2016). Nucleosome positioning: resources and tools online. Briefings in Bioinformatics 17, 745-757.  | Published version | Author’s PDF

** Also see: Experimental nucleosome positioning database | Software to analyse nucleosome positioning experiments |

DescriptionWeb-interfaceLocal installationDi-/tri-nucleotidePeriodicityk-mersEmpirical features
ICM Web: ICM Web allows users to assess nucleosome stability and fold any sequence of DNA into a 3D model of chromatin (Sereda & Bishop, 2010; Stolz & Bishop, 2010). The model is displayed in the visual browser JSmol or can be downloaded. ICM takes a DNA sequence and generates (i) a nucleosome energy level diagram, (ii) coarse-grained representations of free DNA and chromatin and (iii) plots of the helical parameters (Tilt, Roll, Twist, Shift, Slide and Rise) as a function of position. +----+
FineStr: Single-base-resolution nucleosome mapping server (Gabdank et al, 2010; Trifonov, 2010). The analysis is performed using the probe based on the 117-bp DNA bendability matrix derived from C. elegans. The authors suggested the universality of this pattern for other species. +-+++-
iNuc-PhysChem: Identifying nucleosomal or linker sequences from physicochemical properties (Chen et al, 2012). The algorithm identifies nucleosomal sequences by incorporating twelve physicochemical properties defined elsewhere, such as A-philicity, base stacking, B-DNA twist, bendability, bending stiffness, DNA denaturation energy, Z-DNA potential. The model was trained on data from H. sapiens, C. elegans and D. melanogaster. +++-++
iNuc-PseKNC: A sequence-based predictor for nucleosome positioning in genomes with pseudo k-tuple nucleotide composition (Guo et al, 2014). This is another software package from the developers of iNuc-PhysChem. Here, the samples of DNA sequences were formulated using six basic DNA local structural properties trained on datasets from H. sapiens, C. elegans and D. melanogaster. +-+-++
LeNup: Learning Nucleosome positioning from DNA sequences with improved convolutional neural networks. LeNup is a Python based open-source package based on convolutional neural networks to predict nucleosome positioning in H. sapiens, C. elegans, D. melanogaster as well as S. cerevisiae genomes, trained on benchmark datasets. -+--+-
Mapping_CC: Displays the nucleosome predictions based on the DNA dinucleotide correlation pattern. This algorithm was initially associated with one of the first high-throughput genome-wide nucleosome maps in Yeast (Ioshikhes et al, 2006). An updated version is available at -+++--
MOSAICS: Methodologies for Optimization and Sampling in Computational Studies (Minary & Levitt, 2014; Krawczyk, 2018). Perl scripts and a precompiled package to perform training-free atomistic prediction of nucleosome occupancy based on all-atom force field calculations. The effect of DNA methylation can be taken into account. -++--+
NucEnerGen: Nucleosome energetics predictions based on high throughput sequencing (Locke et al, 2010). It utilizes dynamic programming to calculate allowed nucleosome configurations and the Percus equation to infer sequence-dependent energies from the experimental occupancy profiles. -++++-
nuMap: A web application implementing the YR and W/S schemes to predict nucleosome positioning (Alharbi et al, 2014). The methodology is based on the sequence-dependent anisotropic bending, which dictates how DNA is wrapped around a histone octamer. This application allows users to specify a number of options such as schemes and parameters for threading calculation and provides multiple layout formats. +-++-+
NuPoP: Nucleosome Positioning Prediction Engine (Wang et al, 2008; Xi et al, 2010). NuPoP is built upon a duration hidden Markov model, in which the linker DNA length is explicitly modeled. NuPoP outputs the Viterbi prediction, nucleosome occupancy score (from backward and forward algorithms) and nucleosome affinity score. NuPoP has three formats including a web server prediction engine, a stand-alone Fortran program, and an R package. The latter two can predict nucleosome positioning for a DNA sequence of any length.
Nu-OSCAR: Nucleosome-Occupancy Study for Cis-elements Accurate Recognition. It is devoted to identifying binding sites of known transcription factors, which further incorporates nucleosome occupancy around sites on promoter regions. The derivation of the algorithm is based on a biophysical view of interactions between protein factors and nucleosome DNA.
nuScore: A nucleosome-positioning score calculator based on the DNA curvature properties (Tolstorukov et al, 2008). This software allows an important type of analysis, where a user enters many sequences to calculate the average nucleosome energy profile.
N-score: MATLAB and Python codes using a wavelet analysis based model for predicting nucleosome positions from DNA sequence (Yuan & Liu, 2008).
NXSensor: Prediction of nucleosome-excluding sequences based on DNA bending properties (Luykx et al, 2006). It takes as input DNA sequences in FASTA format, and outputs nucleosome-excluding or nucleosome favouring segments.
Segal Lab nucleosome positioning prediction (Field et al, 2008; Kaplan et al, 2009; Segal et al, 2006). This is one of the most popular tools in this class, realized as a web server (allows analyzing a limited number of DNA sequences), and a stand-alone application which can be installed on a local cluster. It allows calculating nucleosome occupancy or nucleosome start site probability profiles of non-overlapping nucleosomes; alternatively, it is possible to calculate the net nucleosome formation energy profile. It uses machine learning for energy assignment based on the training datasets and dynamic programming to sample nucleosome configurations (similar to NucEnerGen, NuPoP and the algorithm of van Noort and co-authors).
Phase: A web server for prediction of the nucleosome formation probability based on (i) the 10-11 bp periodicities of dinucleotides and (ii) the typical pattern "linker-nucleosome-linker" defined by the authors (Levitsky et al, 2014).
RECON: A web server for prediction of the nucleosome formation potential learned from dinucleotide frequencies distribution for nucleosome positioning sequences (Levitsky, 2004; Levitsky et al, 1999).
SymCurv: A program for nucleosome positioning prediction (Nikolaou et al, 2010). It calculates the curvature of the DNA sequence and uses a greedy algorithm to parse the sequence in nucleosome-bound and nucleosome-free segments.
Schiessel Lab nucleosome positioning prediction. This resource contains software packages based on two types of algorithms: nucleosome mutation Monte Carlo (Eslami-Mossallam et al, 2016) and nucleosome positioning with Markov chains (Tompitak et al, 2017). The latter combines the "mutation Monte Carlo" method with dynamic programming similar to NucEnerGen, NuPoP and the algorithm of Segal and co-authors mentioned above.
Trifonov’s strong nucleosomes: Based on the discovery of strong nucleosome positioning sequences which are visually seen as regular arrays in genomic sequence (Nibhani & Trifonov, 2015; Salih et al, 2015), the program from Trifonov’s lab is finding a specific class of strongly positioned nucleosomes of the RR/YY and TA periodic types. +-++--
van Noort Lab nucleosome positioning prediction (van der Heijden et al, 2012). This algorithm is based on dinucleotide distributions, but unlike other methods based on dinucleotide distributions it does not use machine learning and accounts only for the dinucleotide periodicity. In addition, this method uses dynamic programming to account for size exclusion and the Percus equation to assign nucleosome affinities (similar to NucEnerGen, NuPoP and the algorithms of Segal and co-authors and Schiessel and co-authors mentioned above).