# F-Seq2
## Improving the feature density based peak caller with dynamic statistics
Tag sequencing using high-throughput sequencing technologies are employed to identify specific sequence features such as
DNase-seq, ATAC-seq, ChIP-seq, and FAIRE-seq. To intuitively summarize and display individual sequence data as an
accurate and interpretable signal, we have developed the original [F-Seq](http://fureylab.web.unc.edu/software/fseq/)
[GitHub](https://github.com/aboyle/F-seq), a software package that generates a continuous tag sequence density
estimation allowing identification of biologically meaningful sites whose output can be displayed directly in the UCSC
Genome Browser.
F-Seq2 is a complete rewrite of the original version in Python. We designed a new statistical framework and introduced
new features to F-Seq to further improve the performance in its second version. F-Seq2 implements a dynamic
parameter to conduct local statistical analysis with an underlying “continuous” Poisson distribution. By combining the
power of the local test and the KDE, which model the read probability distribution with statistical rigor, we robustly
account for local biases and solve ties that occur when ranking candidate summits, making results suitable for
irreproducible discovery rate (IDR) analysis.
## Table of contents
1. [Installation](./INSTALL.md)
2. [Usage](#usage)
- [`callpeak`](#callpeak)
- [`callpeak_idr`](#callpeak_idr)
- [`idr`](#idr)
3. [Output files and formats](#output-files-and-formats)
4. [Examples](#examples)
5. [Reference](#reference)
6. [Troubleshooting](#troubleshooting)
## Installation
Prerequisite: [BEDTools](https://bedtools.readthedocs.io/en/latest/content/installation.html).
See [here](./INSTALL.md) for more details to install F-Seq2.
## Usage
```
fseq2 [-h] [--version]
{callpeak, callpeak_idr, idr}
```
Available subcommands
Subcommand | Description
-----------|----------
`callpeak` | F-Seq2 main function to call peaks from alignment results.
`callpeak_idr` | Call peaks and follow by IDR framework with recommended settings.
`idr` | A wrapper for [IDR package](https://github.com/nboley/idr) for customized IDR analysis.
## `callpeak`
#### Command line input:
##### `-treatment_file`
REQUIRED argument for fseq2. Treatment file(s) in bam or bed format. If specifiy multiple files (separated by space),
they are considered as one treatment experiment. See [here](./INPUT_FORMAT.md) for more details about input format.
##### `-control_file`
Control file(s) corresponding to treatment file(s).
##### `-pe`
Paired-end mode. If this flag on, treatment (and control) file(s) are paired-end data, either in format of BAMPE or BEDPE.
Default is False to treat all data as single-end. See [here](./INPUT_FORMAT.md) for more details about paired-end mode.
##### `-chrom_size_file`
A file specify chrom sizes, where each line has one chrom and its size. This is required if output signal format is bigwig.
Note if this file is specified, fseq2 only process the chroms in this file. Default is False to process all and cannot output bigwig.
##### `-o`
Output directory. Default is current directory.
##### `-name`
Prefix for all output files. This overrides exisiting files. Default is `fseq2_result`.
##### `-sig_format`
Signal format for reconstructed signal. Available format `wig`, `bigwig`, `np_array`. Note if choose `np_array`, arrays
for each chrom are stored in [`NAME_sig.h5`](#name_sigh5) with `chrom` as key, and no gaussian smooth applied. Default is False, without output signal.
##### `-sort_by`
Sort peaks and summits by `pValue` or `chromAndStart`. Default is `chromAndStart`.
##### `-v`
Verbose output. Default is False.
##### `-f`
Fragment size of treatment data. Default is to estimate from data. This determines shift size where `offset = fragment_size/2`.
For DNase-seq and ATAC-seq data, set `-f 0`.
##### `-l`
Feature length for treatment data. Default is 600. Recommend 50 for TF ChIP-seq, 600 for DNase-seq and ATAC-seq,
1000 for histone ChIP-seq.
##### `-fc`
Fragment size of control data.
##### `-t`
Threshold (standard deviations) to call candidate summits. Default is 4.0. Recommend 4.0 for broad peaks,
8.0 for sharp peaks.
##### `-p_thr`
P value threshold. Default is 0.01. Consider to relax it to 0.05 when without control data or calling broad peaks.
##### `-q_thr`
Q value (FDR) threshold. Default is not set and use `p_thr`. If set, only use `q_thr`.
##### `-cpus`
Number of cpus to use. Default is 1.
##### `-tp`
Threshold (standard deviations) to call peak regions. Default is 4.0.
##### `-sparse_data`
If flag on, statistical test includes 1k region for more accurate background estimation. This can be useful for single-cell data.
##### `-nfr_upper_limit`
Nucleosome free region upper limit. Default is 150. Used as window_size and min_distance when `-f 0`.
##### `-pe_fragment_size_range`
Effective only if `-pe` on. Only keep PE fragments whose size within the range to call peaks. Default is False,
without any selection. Useful for ATAC-seq data:
(1) to call peaks on nucleosome free regions, specify: `0 150`
(2) to call peaks on nucleosome centers, specify: `150 inf`
(3) to call peaks on open chromatin regions, specify: `auto`
> `auto` is a filter designed for ATAC-seq open chromatin peak calling where we filter out fragments whose size related to
mono-, di-, tri-, and multi-nucleosomes. Size information is taken from the original ATAC-seq paper (Buenrostro et al.).
You can design your own auto filter based on specific experiment data by specifying `-nucleosome_size` parameter.
##### `-nucleosome_size`
Effective only if `-pe` on and specify `-pe_fragment_size_range auto`. Default is `180, 247, 315, 473, 558, 615` They
are the ATAC-seq PE fragment sizes related to mono-, di-, and tri-nucleosomes. Fragments whose size within the ranges
and above the largest bound (i.e. 615) are filtered out when calling peaks. Change those numbers to design your own auto filter.
##### `-prior_pad_summit`
Prior knowledge about peak length which only padded into `NAME_summits.narrowPeak`. Default is 0.
Useful for IDR analysis: in `callpeak_idr`, we set it to est. fragment size.
##### `-num_peaks`
Maximum number of peaks called. Default is not set. If set, overrides `p_thr` and `q_thr`.
## `callpeak_idr`
#### Command line input:
Most arguments are shared between `callpeak` and `callpeak_idr`. Here are the unique ones.
> Notice if it is `-` or `--` ahead of arguments. `--` arguments are from IDR package. `-` are from fseq2.
##### `-treatment_file_1`
Treatment file in bam or bed format as replicate 1.
##### `-treatment_file_2`
Treatment file in bam or bed format as replicate 2.
##### `-control_file_1`
Control file in bam or bed format, paired with replicate 1 treament file.
##### `-control_file_2`
Control file in bam or bed format, paired with replicate 2 treament file.
##### `-name_1`
Prefix for output files for replicate 1 (default=`fseq2_result_1`).
##### `-name_2`
Prefix for output files for replicate 2 (default=`fseq2_result_2`).
##### `-prior_pad_summit`
Prior knowledge about peak length which only padded into `NAME_summits.narrowPeak`. Default is est. fragment size.
##### `--idr_threshold`
Only return peaks with a global idr threshold below this value. Default: report all peaks.
##### `--soft_idr_threshold`
Report statistics for peaks with a global idr below this value but return all peaks with an idr below --idr Default: 0.05.
##### `--plot`
Plot IDR results. Specify False if no plot. Default is to plot to `NAME_1_NAME_2.png`. Can specify other name here.
Notice this is different from original IDR package which is only a flag.
## `idr`
#### Command line input and output:
See original [IDR documentation](https://github.com/nboley/idr#usage).
> Notice all single letter arguments are removed to avoid conflict with fseq2, e.g. no `-s`, use `--samples`
## Output files a
PyPI 官网下载 | fseq2-2.0.0.tar.gz
版权申诉
196 浏览量
2022-01-11
21:29:38
上传
评论
收藏 71KB GZ 举报
挣扎的蓝藻
- 粉丝: 13w+
- 资源: 15万+
最新资源
- n.cpp
- jdk-8u411-windows-x64下载安装可用
- vgg模型-图像分类算法对水果识别-不含数据集图片-含逐行注释和说明文档.zip
- KMP算法(Knuth-Morris-Pratt算法
- vgg模型-python语言pytorch框架训练识别化妆品分类-不含数据集图片-含逐行注释和说明文档.zip
- KMP算法(Knuth-Morris-Pratt算法
- shufflenet模型-基于人工智能的卷积网络训练识别狗的表情-不含数据集图片-含逐行注释和说明文档.zip
- shufflenet模型-python语言pytorch框架训练识别张嘴闭嘴-不含数据集图片-含逐行注释和说明文档.zip
- resnet模型-基于人工智能的卷积网络训练识别面部表情识别-不含数据集图片-含逐行注释和说明文档
- resnet模型-python语言pytorch框架训练识别香蕉品质-不含数据集图片-含逐行注释和说明文档.zip
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈