NucHMM: a method for quantitative modeling of nucleosome organization identifying functional nucleosome states distinctly associated with splicing potentiality.

Introduction

Nucleosome organization, described as nucleosomal phasing, spacing, and positioning, is determined by the interplay among nucleosome, nucleosome-binding factors such as DNA-binding factors, histone chaperones, and ATP-dependent chromatin remodelers. To address the lacking power of determining the combinational effects of the different influencing factors on nucleosome organization, we presented NucHMM for identifying functional nucleosome states. NucHMM integrates a hidden Markov model (HMM), and estimated nucleosome phasing, spacing as well as positioning information, to identify nucleosome states associated with cell type-specific combinatorial histone marks.

NucHMM Update Logs (version 1.3)

v.1.3.2 08/21/2023

fixed bug for write_array function and change the spacing of 'single' type to -1

v.1.3.1 02/20/2023

fixed bugs for segment_gene function
add hg38 gene reference file

v1.3 01/09/2023

polished codes to make runing more smoothly
fixed several bugs

v1.2 02/21/2022

(important) Optimize and fix bug for the installation step

v1.2

add --ciselementlist parameter in nuchmm-init
optimize the nucleosome organization filters and the final output format
fix typo in NucHMM_common_data.py

v1.1

add --markthreshold parameter in nuchmm-screen-init and matrix-visualize
fix background_state bug for nuchmm-screen-init

Workflow

Installation

The program is tested on Linux 3.10.0/CentOS 7.

NucHMM was implemented by using python. It has been compiled and run exclusively on Linux operating systems. The C++ part of NucHMM should compile with GCC and run on any x86 64-bit Linux system running at least Kernel 2.6. A computer with several gigabytes of RAM is strongly recommended to avoid an out-of-memory error. NucHMM requires Boost and OpenMP to be available (can be install by conda). By default, NucHMM will attempt to use all cores of a computer if possible (up to a limit depending on the genome and the number of cell lines used, after which additional cores will provide no benefit), however NucHMM respects the environment variable OMP_NUM_THREADS. NucHMM does not benefit from any GPU installed or CUDA, and is CPU-bound instead of IO-bound on most systems.

The installation currently requires the miniconda.

Install miniconda:

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh

Download NucHMM:

git clone https://github.com/KunFang93/NucHMM.git

Create the conda environment with dependents and Activate the environment:

conda env create -n nuchmm -f <path to NucHMM>/scripts/env.yaml
conda activate nuchmm

Install NucHMM:

cd <path to the NucHMM>/scripts
pip install --editable .

Quick Start

Note: The context in <> should be replace by user according to real data

Step1(Optional): ChIP-seq peaks and MNase-seq nucleosome locations files preparation.

Skip this step if you alreay have the histone marks' peak files and iNPS derived nucleosome files (remove the header, and please keep the all like_wig files). You can check those files' format in example_files folder. We recommend use your own favored pipeline.

This section needs additional installation of packages

use conda install
  - bowtie=1.2.3
  - bowtie2=2.4.1
  - fastqc=0.11.9
  - trim-galore=0.6.5
  - macs2=2.2.7.1
  - epic2=0.0.41

If the input all files in fastq format

# not include MNase-seq fq
NucHMM nuchmm-prep --fastq -p 20 -ifql fqs_list.txt -qc -bip <Full Path to bowtie index>/<idx-basename>

# include MNase-seq fq
NucHMM nuchmm-prep --fastq -p 20 -ifql fqs_list.txt -qc -bip <Full Path to bowtie index>/<idx-basename> -inps <Full Path to NucHMM>/scripts/iNPS_V1.2.2.py

If the input all files in bam format

# not include MNase-seq bam
NucHMM nuchmm-prep --bam -p 20 -ibl bams_list.txt

# include MNase-seq bam
NucHMM nuchmm-prep --bam -p 20 -ibl bams_list.txt -inps <Full Path to NucHMM>/scripts/iNPS_V1.2.2.py

The output peak files will locate at peakcalling_result/ folder, and the output nucleosome location files will locate at nuc_calling_result/ folder.

Step2: NucHMM initialization.

First manually create

# Check the those files format in exmaple_files folder.
1. <celltype>_histone_marks.txt that contains all histone mark peak files;
2. histonelists_list.txt contains all <celltype>_histone_marks.txt file;
3. nucposfiles_list.txt contains all <celltype>_nucleosome_locations.bed.

Then, use command

NucHMM nuchmm-init -iplf histonelists_list.txt -nucf nucposfile_list.txt -gf <Full Path to NucHMM>/annotation/genebody_anno_hg19.txt -rmf

e.g.

NucHMM nuchmm-init -iplf histonelists_list.txt -nucf nucposfile_list.txt -gf /data/kun/Softwares/NucHMM/annotation_files/genebody_anno_hg19.txt -rmf

The default output file will be named _<# histone mark>.precomp in the current directory. User can specify the ouptut name by -ofl parameter.

Step3: NucHMM training.

First manually create.

precompfiles_list.txt that contains all precomp files result from nuchmm-init. 
(check the file format in example_files/histone_marks.txt)

Then,

NucHMM --hmm-directory <Full Path to NucHMM>/scripts/NucHMM_Cplus/bin/ nuchmm-train -refg <Full Path to NucHMM>/annotation/hg19.chrom.sizes.txt -pl precompfile_list.txt -numh <number of histone marks. e.g. 8> -nums <number of input states. e.g. 20>

Note that training process could take few hours up to a day depends on how much data you input. We normally use nohup command to run the this nuchmm-train command in the background by

nohup NucHMM --hmm-directory <Full Path to NucHMM>/scripts/NucHMM_Cplus/bin/ nuchmm-train -refg <Full Path to NucHMM>/annotation/hg19.chrom.sizes.txt -pl precompfile_list.txt -numh <number of histone marks. e.g. 8> &> train.log &

e.g.

nohup NucHMM --hmm-directory /data/kun/Softwares/NucHMM/scripts/NucHMM_Cplus/bin nuchmm-train -refg /data/kun/NucHMM_project/NucHMM_data/annotation/hg19.chrom.sizes.txt -pl /data/kun/project/nuchmm_results/nuchmm_init/Precompfiles_list.txt -numh 8 -nums 15 -rmf &>train_15_1.log &

The default output file will be name HMM_<# histone marks>.rawhmm in the current directory. User can specify the output name by -ohmm parameter.

Step4: Matrix visualization and background state detection.

Check if there is a auto-generated histone_marks.txt file(Generated in NucHMM-init process), it not, then manually create

histone_marks.txt file that contains all histone marks. 
(check the file format in example_files/histone_marks.txt)

Then, visualize the Transition and Mark-state matrix first.

NucHMM matrix-visualize -rhf HMM_<# histone marks>.rawhmm -hlf histone_marks.txt

e.g.,

NucHMM matrix-visualize -rhf /data/kun/project/nuchmm_results/nuchmm_train/18/18_2/HMM_8_secondr.rawhmm -hlf histone_marks.txt -tmat trans_18-2.png -msmat markstate_18-2.png -mt 0.3

The default output files will be two png files.

View the Mark_state matrix and define no histone mark HMM states as background state. 
(e.g. in example_files/Mark_state.png, 8 and 12 is background states)

Step5: NucHMM screen initialization.

Note the order of parameter should be same with the cell type order in histonelists_list.txt

NucHMM nuchmm-screen-init -rhf HMM_<# histone marks>.rawhmm -hlf histone_marks.txt \
-bg <background state1> -bg <background state2> .. -bg <background stateN> \
-gf <Full Path to NucHMM>/annotation/genebody_anno_hg19.txt \
-ct <celltype1> -ct <celltype2> .. -ct <celltypeN> -ptm -rmf

e.g.,

NucHMM nuchmm-screen-init -rhf /data/kun/project/NucHMM_results/NucHMM_train/110920/20/20_1/HMM_8_secondr.rawhmm -hlf /data/kun/project/NucHMM_results/NucHMM_init/110920/histone_marks -bg 1 -bg 2 -gf /data/kun/NucHMM_project/NucHMM_data/annotation/genebody_anno.txt -ct Abl -ct LNCap-Veh -ptm -rmf -sfl statesfilelist.txt  -ofl outputfilelist.txt

The default output files:
Total_distribution_up.png, Total_distribution_gbdown.png, Total_distribution.png
nuchmm_screen_init_result_files.txt
states_genomic_location.txt.

Step6: NucHMM screen.

Firstly,

Double-Check states_genomic_location.txt and Total_distribution.png.
Make sure the predicted genomic location in states_genomic_location.txt fits your empirical knowledge.
If not, you can manually edit the states_genomic_location.txt file.
create like_wigs_list.txt

Then,

NucHMM nuchmm-screen -gf <Full Path to NucHMM>/annotation/genebody_anno_hg19.txt \
-lwfl like_wigs_list.txt -nucf nucposfiles_list.txt -bg <background state1> .. -bg <background stateN> \
-sn <Total number of states, e.g. Mark_state.png show total 13 states> -ifl nuchmm_screen_init_result_files.txt \
 -srf states_genomic_location.txt -pm -wi -rmf

e.g.,

NucHMM nuchmm-screen -gf /data/kun/NucHMM_project/NucHMM_data/annotation/genebody_anno.txt -lwfl like_wigs_list.txt -nucf nucposfile_list.txt -bg 3 -bg 13 -sn 14 -pm -wi -rmf -ifl /data/kun/project/nuchmm_results/nuchmm_screen/nuchmm_screen_init_result_files.txt -srf /data/kun/project/nuchmm_results/nuchmm_screen_init/23_3/states_genomic_location.final.txt

The default output files:

<celltypes>_gl_an_resp_pos_final.bed
<celltypes>_gl_an_resp_pos_final.arrayed.bed
functional_nucleosome_state_post.txt
other temporary files and figures are stored in detail_info folder

The final function nucleosome states features are store in functional_nucleosome_state_post.txt.
The final kept and annotated nucleosomes are in < celltypes >_gl_an_resp_pos_final.bed. The columns in < celltypes >_gl_an_resp_pos_final.bed are:

Chrom   Start   End State   Local.Positioning.Score Local.Spacing   Local.Phasing.info  Positioning.Mark

The final simplified nucleosome states' array are in < celltypes >_gl_an_resp_pos_final.arrayed.bed.
The columns in < celltypes >_gl_an_resp_pos_final.arrayed.bed are:

Chrom   Start   End State   Local.Ave.Positioning   Local.Ave.Spacing

Enjoy 😊!

Command Line

Usage

NucHMM [OPTIONS] COMMAND [ARGS]

Options:
  --version               Show the version and exit
  --hmm-directory PATH    the path of the NucHMM_Cplus/bin folder
  --help                  Show this message and exit

Commands:
  nuchmm-prep             Prepare ChIP-seq peaks files or nucleosome locations files from fastq/bam files.
  nuchmm-init             Assign histone marks to nucleosomes and create precomp bins for nuchmm-train.
  nuchmm-train            Use Hidden Markov Model(HMM) and Viterbi Algorithm to decode HMM states for each nucleosomes.
  nuchmm-screen-init      Initializae the screen step by providing sorted state files and the suggested genomic locations of HMM states.
  nuchmm-screen           Filter nucleosomes by genomic location, array number, nucleosome regularity, spacing and positioning.
  matrix-visualize        Visualize the Transition and Mark-state matrix.

Options
- --version:
  Show NucHMM current version and exit.
- --hmm-directory PATH:
  Used in nuchmm-train for input the path of the NucHMM_Cplus/bin folder.
- --help:
  Show help message.
Commands
- nuchmm-prep
  
  NucHMM provides basic ChIP-seq and MNase-seq pipeline to handle the fastq/bam files. If you have other favored ChIP-seq pipeline, we recommend use your favored way to process the raw fastq or bam files. However, for MNase-seq, we only accept result from iNPS currently.
  - --fastq/--bam:
    Indicate the input data type
  - --inputfqslist(-ifql):
    Used with --fastq command. Input the fastqs list file. Input format: 'FILE READ-LEN SEQ-TYPE(chip/mnase) PEAK-TYPE(narrow?broad/none)'. Check fqs_list.txt in example_files folder for the detailed format.
  - --inputbamslist(-ibl):
    Used with --bam command. Input the bams list. Input format: for each line 'FILE PE?SE SEQ-TYPE(chip/mnase) PEAK-TYPE(narrow?broad/none)'. Check bams_list.txt in example_files folder for the detailed format.
  - --qualitycontrol(-qc):
    Used with --fastq command. Flag for whether run QC trim step for fastq files.
  - --bowtieindexpath(-bip):
    Required if input format is fastq, the path and basename of bowtie index (bowtie -x)
  - --bowtie2indexpath(-b2ip):
    Required if input format is fastq, the path and basename of bowtie2 index (bowtie2 -x)
  - --inpspath(-inps):
    Required if input has MNase-seq, the path of iNPS.py. For example, /data/NucHMM/scripts/iNPS_V1.2.2.py.
  - --threads(-p):
    Number of threads
- nuchmm-init
  
  Assign histone marks to nucleosomes and create precomp bins for nuchmm-train. All .precomp files name are writing to Precompfiles_list.txt.
  - --inputpeakslistfiles(-iplf)[Required]:
    Input TF-peaks_files list. Check the histonelists_list.txt in example_files folder for the detailed format.
  - --nucpositionfiles(-nucf)[Required]:
    Input the nucleosome position files list. Check the nucposfiles_list.txt in example_files folder for the detailed format.
  - --genefile(-gf)[Required]:
    Input the interested genes list file. Check the genebody_anno_hg19.txt in annotation_files for the detailed format.
  - --outputfilelist(-ofl):
    Specify the output files path and name.
    Format:
    < Path >/< celltype1 >.precomp
    < Path >/< celltype2 >.precomp
    ..
    < Path >/< celltypeN >.precomp
  - --refgenome(-refg):
    The reference genome file contains the length of each chromosomes. Default: built-in hg19.
  - --intersect_cutoff(-ic):
    The intersect threshold to assign histone mark peaks to the nucleosomes. Default is 0.3. For example, for a 150 bp nucleosome core region, if a histone mark peak have 45 (150 * 0.3) bp overlap with this nucleosome. Then the program will consider this nucleosome has this histone mark.
  - --gap(-g):
    Flag of giving a mark to inter-nucleosome region (gaps between nucleosomes).
  - --ciselementlist(-cel): The list of cell cis-element files. Allow users to add the cis-element regulators in the HMM model. Check the example file in example_files folder. Warning: This comment will conflict with nuchmm-screen-init and nuchmm-screen! Only applied on nuchmm-train, if user want to perform the following nucleosome organization analysis, please remove the gaps in the bed file from nuchmm-train.
  - --upboundary(-up):
    Upstream boundary of TSS for selecting the training region. Default:100000. The larger the number, the more computational resources (memory) and training time needed.
  - --downboundary(-down):
    Downstream boundary of TTS for selecting the training region. Default:10000. The larger the number, the more computational resources (memory) and training time needed.
  - --removetmpfile(-rmf): Flag of removing all temporary files.
- nuchmm-train
  
  Use Hidden Markov Model(HMM) and Viterbi Algorithm to decode HMM states for each nucleosomes.
  - --refgenome(-refg)[Required]:
    Input the reference genome file that contains the length of each chromosomes. Check the hg19.chrom.sizes.txt in annotation_files folder.
  - --precomp_list(-pl)[Required]:
    Input precomp files list resulted from nuchmm-init. Check the precompfiles_list.txt in example_files folder.
  - --numhistonemarks(-numh)[Required]:
    The number of histone marks used. For example, if user use H3K4me1/K4me3/K27ac in nuchmm-init, then input 3.
  - --numStates(-nums):
    The number of states for first round hmm training. Default: 20.
  - -b:
    Calculate and report BIC after every iteration. WARNING: This parameter will severely slow the algorithm!
  - --bicfile(-obic):
    Specify the path and name of the BIC score file. If -b not specified, only report BIC score of the first and the last iteration.
  - -i:
    Maximum number of iterations for first round hmm training. Default: 300.
  - --num/--perc:
    The crietrion to remove the redundant states in the first round hmm training. --num uses the specific number as cutoff while --perc uses a specific percentage as cutoff. Default use is --perc (0.005). If specify --num, then default number is 350. For example, --num -nn 300 function as: if a HMM state has less than 300 nucleosomes in the first round hmm training, then this HMM state will be removed from the tranisition and emission matrix which are used as the initial matrix in the second round HMM. --perc -pn 0.01 function as: if a HMM state has less than 0.01% of the total number of nucleosomes, then this HMM will be removed from the tranisition and emission matrix which are used as the initial matrix in the second round HMM.
  - --nucnum(-nn):
    Used with --num to specify the number cutoff.
  - --percentage(-pn): Used with --perc to specify the percentage cutoff.
  - --outrawhmmfile(-ohmm):
    Specify the path and name of the output rawhmm file.
  - --outstats(-os):
    Output the nucleosome counts for each HMM states in each cell type.
  - --removetmpfile(-rmf):
    Remove temporary files.
- nuchmm-screen-init
  
  Initializae the screen step by providing sorted state files and the suggested genomic locations of HMM states.
  - --rawhmmfile(-rhf)[Required]:
    Input the secondr.rawhmm file resulted from nuchmm-train.
  - --histonelistfile(-hlf)[Required]:
    Input histone_marks.txt file that contains all histone marks used in nuchmm-init. Check the histone_marks.txt for the detail format.
  - --genesfile(-gf)[Required]:
    Input the interested genes list file. Check the genebody_anno_hg19.txt in annotation_files for the detailed format.
  - --celltypes(-ct)[Required]:
    Input all cell types name used in nuchmm-init. For example, if user train model with data from MCF7, H1 and IMR90, user should input -ct MCF7 -ct H1 -ct IMR90. The order of the -ct should be the same with the celltype order in nucposfiles_list.txt.
  - --bgstate(-bg)[Required if exists background states]:
    Specify background states that identified from the Mark-state matrix. If there are multiple background states, for example HMM state1 and HMM state2 are background states, then use -bg 1 -bg 2.
  - --statesfilelist(-sfl):
    Specify the path and name of the output < celltype > _states_secondr.bed. The order of the < celltype > should be same with -ct command. The command will detect the states_secondr.bed in current directory automatically. If user want to run command in the background, this parameter is required.
    format:
    < celltype1 > _states_secondr.bed
    ..
    < celltypeN > _states_secondr.bed
  - --outputsfilelist(-ofl):
    Specify the path and name of the output < celltype > _output_secondr.bed. The order of the < celltype > should be same with -ct command. The command will detect the output_secondr.bed in current directory automatically. If user want to run command in the background, this parameter is required.
    format:
    < celltype1 > _output_secondr.bed
    ..
    < celltypeN > _output_secondr.bed
  - --plotcellmark(-pcm):
    Plot the cell type specific HMM states distribution.
  - --plottotalmark(-ptm):
    Plot the all cell averaged HMM states distribution.
  - --overwrite(-ow):
    Overwrite all existing files. Useful when user need to re-run command in the same directory.
  - --winmethod/--refmethod:
    Pick a method to identify unique HMM state for the nucleosomes that have multi-states. --winmethod uses window method to identify the unique state; --refmethod uses output_secondr.bed as reference to identify the unique state. Default method: --refmethod. See the pick_one function in NucHMM_Feature_screen for the detailed inplementation.
  - --winsize(-ws):
    Used with --winmethod, Default value: 5. Window method use the adjcent nucleosome states in the selected window to identify the unique state for the target nucleosome. For example, if a nucleosome has both HMM state1(associated with H3K4me1) and HMM state2(associated with H3K4me3) labelled, we then select the upstream < 5 > nucleosomes and downstream < 5 > nucleosomes of this target nucleosome to find which HMM state is most likely appeared under this environment. See the pick_one function in NucHMM_Feature_screen for the detailed inplementation.
  - --updistal(-uD):
    Specify upstream boundary for plotting the distribution, cannot exceed than --upboundary used in nuchmm-init. Default: 100000.
  - --upproximal(-uProx):
    Specify upstream boundary of the proximal region. Used in identifying the genomic location of HMM states. Default: 5000.
  - --uppromoter(-uProm):
    Specify upstream boundary of the promoter region. Used in identifying the genomic location of HMM states. Default: 1000.
  - --downbound(-db):
    Specify downstream boundary for plotting the distribution, cannot exceed than --downboundary used in nuchmm-init. Default: 10000.
  - --rescalelength(-rl):
    Specify the rescaled gene body length. Default: 30000.
  - --markthreshold(-mt):
    Specify the mark threshold for state-mark matrix. Default: 0.25.
  - --refhg19/--refhg38:
    Specify the reference genome. Default: built-in hg19.
  - --outfile(-of):
    Specify the path and name of the identified genomic location file.
  - --removetmpfile(-rmf):
    Remove temporary files.
- nuchmm-screen
  
  Filter nucleosomes by genomic location, array number, nucleosome regularity, spacing and positioning.
  - --genesfile(-gf)[Required]:
    Input the interested genes list file. Check the genebody_anno_hg19.txt in annotation_files for the detailed format.
  - --like_wig_fileslist(-lwfl)[Required]:
    Input the like_wigs_list.txt that contains < celltype > _like_wigs.txt. Check the like_wig_list.txt in example_files folder for the detail format and information. The order of the cell types should be the same with the celltype order in nucposfiles_list.txt.
  - --nucpositionfiles(-nucf)[Required]:
    Input the nucleosome position files list. Check the nucposfiles_list.txt in example_files folder for the detailed format.
  - --statesnumber(-sn)[Required]:
    Input the total number of HMM states (Include background states.)
  - --bgstate(-bg)[Required if exists background states]:
    Specify background states that identified from the Mark-state matrix. If there are multiple background states, for example HMM state1 and HMM state2 are background states, then use -bg 1 -bg 2.
  - --inputfileslist(-ifl):
    Input the list of files resulted from nuchmm-screen-init. Check nuchmm_screen_init_result_files.txt in example_files folder for the detailed information. The program will automatically look for the nuchmm_screen_init_result_files.txt in current directory.
  - --statesregionfile(-srf):
    Input the states_genomic_location.txt (contains the predicted genomic location for each non background HMM states) resulted from nuchmm-screen-init. Check states_genomic_location.txt in example_files folder for the detailed information. The program will automatically look for the states_genomic_location.txt in current directory.
  - --updistal(-uD):
    Specify upstream boundary for genomic location filtering, cannot exceed than --upboundary used in nuchmm-init. Default: 100000.
  - --upproximal(-uProx):
    Specify upstream boundary of the proximal region for genomic location filtering. Default: 5000.
  - --uppromoter(-uProm):
    Specify upstream boundary of the promoter region for genomic location filtering. Default: 1000.
  - --cutoffdist(-cfd):
    Specify the maximum distance between two nucleosomes in a nucleosome array. If the distance between two nucleosome larger than this cutoff distance, the program will consider these two nucleosomes are in different nucleosome arrays. Default: 350.
  - --upratio(-ur):
    Specify the top quantile cutoff in array-num and nucleosome positioning filtering process. Default: 95.
  - --downratio(-dr):
    Specify the bottom quantile cutoff in array-num and nucleosoem positioning filtering process. Default: 5.
  - --arraydown(-adwon):
    Specify the range of the nucleosome array for calculating nucleosoem regularity and spacing. Currently only accept 1000 and 2000 as input. Default: 1000.
  - --max/--mean:
    Specify the criterion to calculate the regularity score. --max method chooses the max spectral density between 150-200 bp as the regularity score. --mean method calculate the average of teh spectral density between 150-200 bp as the regularity score.
  - --rankcoef(-rc):
    Specify the cofficient of the regularity rank for filtering unmatched nucleosome. The larger the coefficient, the looser the filter. Default: 1.
  - --refhg19/--refhg38:
    Specify the reference genome. Default: built-in hg19.
  - --writeinfo(-wi): Write files contain the detail array-num, nucleosome regularity, spacing and positioning information.
  - --plotmark(-pm):
    Plot mark for ploting state-array distribution, bar-plot of regularity score, spectral-density plot and nuc-pos violin plot.
  - --removetmpfile(-rmf):
    Remove temporary files.
- matrix-visualize
  
  Visualize the Transition and Mark-state matrix.
  - --rawhmmfile(-rhf)[Required]:
    Input the secondr.rawhmm file resulted from nuchmm-train.
  - --histonelistfile(-hlf)[Required]:
    Input histone_marks.txt file that contains all histone marks used in nuchmm-init. Check the histone_marks.txt for the detail format.
  - --matrixcolor(-mc):
    Specify the color palette of the matrix. 0 is red-white; 1 is red-yellow(YlOrRd) and 2 is red-blue(coolwarm). Default: 2.
  - --markthreshold(-mt):
    Specify the mark threshold for state-mark matrix. Default: 1.1 (not showing).
  - --transmat(-tmat):
    Specify the path and name of the transition probability matrix, otherwise will automatically save to trans.< current time >.png
  - --markstatemat(-msmat):
    Specify the path and name of the mark-state matrix, otherwise will automatically save to mark_state.< current time >.png
  - --emitmat(-emat): Specify the path and name of the emission probability matrx. Default: None.
  - --M:
    Divide the element of emission probability matrix by the number of the mark in that observation. Very strong assumption. Not recommend use.
  - --writematrix:
    Write mark_state matrix and transistion probability matrix in txt files.
  - --inputmatrix: Visualize input mark_state matrix

Citations

NucHMM is now published in Genome Biology. If it helps your research, please cite it as:

     Fang, K., Li, T., Huang, Y. et al. NucHMM: a method for quantitative modeling of nucleosome organization identifying functional nucleosome states distinctly associated with splicing potentiality. Genome Biol 22, 250 (2021). https://doi.org/10.1186/s13059-021-02465-1

Contact Us

Kun Fang: [email protected]; Victor Jin: [email protected]

kunfang93 / nuchmm Goto Github PK

nuchmm's Introduction

NucHMM: a method for quantitative modeling of nucleosome organization identifying functional nucleosome states distinctly associated with splicing potentiality.

Introduction

NucHMM Update Logs (version 1.3)

Workflow

Installation

Quick Start

Step1(Optional): ChIP-seq peaks and MNase-seq nucleosome locations files preparation.

Step2: NucHMM initialization.

Step3: NucHMM training.

Step4: Matrix visualization and background state detection.

Step5: NucHMM screen initialization.

Step6: NucHMM screen.

Command Line

Usage

Options

Commands

nuchmm-prep

nuchmm-init

nuchmm-train

nuchmm-screen-init

nuchmm-screen

matrix-visualize

Citations

Contact Us

nuchmm's People

Contributors

Watchers

Forkers

nuchmm's Issues

Recommend Projects

Recommend Topics

Recommend Org