{% extends 'templates/global/base.md' %}

{% block content %}

# Methods

## Experiment Introduction

Each step in the experimental operation (such as sample detection, library construction to sequencing) will affect the quantity and quality of data, thus affecting the follow-up analysis results. In order to obtain high-quality sequencing data, we perform strict quality control in each step. The experiment pipeline is shown in {{data.RenderAnchorLink("figure", "Figure", fn)}}.

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.RenderExpFlowUrl("en")}}
subTitle: Figure{{fn}} Experimental workflow
```

{{ data.RenderExpStep("en")}}

{% set fn = fn + 1 %}

## Bioinformatics Analysis Process

{% if data.ProjectInfo.ProjectType == 'lncRNA' or data.ProjectInfo.ProjectType == 'Hiseq_lncRNA' %}

The data obtained by sequencing is called raw reads. Firstly,  reads with low quality, adapter contamination, and highly unknown N base are filtered. The filtered data is called clean reads. Then the clean reads are aligned to the reference genome, followed by the transcriptions assembly. For new transcripts, we will perform coding ability prediction to distinguish between novel lncRNA and mRNA. These novel lncRNAs and mRNAs will be analyzed with known lncRNAs and mRNAs, subsequently.

We will perform the RNA quantitation, the difference analysis, cluster analysis, time series analysis, enrichment analysis, etc. We will also predict the target genes of lncRNA, predict the lncRNA family classification and potential miRNA precursors. For mRNA, we will perform database annotations. Our analysis also includes alternative splicing and fusion Gene. Thebioinformatics analysis pipelineis as follows {{data.RenderAnchorLink("figure", "Figure", fn)}}.

{% endif %}

{% if data.ProjectInfo.ProjectType == 'DNBSEQ_RNAdenovo' or data.ProjectInfo.ProjectType == 'DNBSEQ_Strand-specific_RNAdenovo' %}

The data obtained by sequencing is called raw reads. Firstly,  reads with low quality, adapter contamination, and highly unknown N base are filtered. The filtered data is called clean reads. Then the clean reads are used to transcriptions assembly. For new transcripts, we will perform coding ability prediction to distinguish between novel lncRNA and mRNA. These novel lncRNAs and mRNAs will be analyzed with known lncRNAs and mRNAs, subsequently.

We will perform the RNA quantitation, the difference analysis, cluster analysis, time series analysis, enrichment analysis, etc. For mRNA, we will perform database annotations. The databases include NR, NT, GO, KOG, KEGG, UniProt, COG, Rfam, Pfam, TF, STRING, etc. Thebioinformatics analysis pipelineis as follows {{data.RenderAnchorLink("figure", "Figure", fn)}}.

{% endif %}


{% if data.ProjectInfo.ProjectType == 'DNBSEQ_RNAref' or data.ProjectInfo.ProjectType == 'DNBSEQ_Strand-specific_RNAref' %}

The data obtained by sequencing is called raw reads. Firstly,  reads with low quality, adapter contamination, and highly unknown N base are filtered. The filtered data is called clean reads. Then the clean reads are aligned to the reference genome, followed by the transcriptions assembly. For new transcripts, we will perform coding ability prediction to distinguish between novel lncRNA and mRNA. These novel lncRNAs and mRNAs will be analyzed with known lncRNAs and mRNAs, subsequently.

We will perform the RNA quantitation, the difference analysis, cluster analysis, time series analysis, enrichment analysis, etc. For mRNA, we will perform database annotations. The databases include NR, NT, GO, KOG, KEGG, UniProt, COG, Rfam, Pfam, TF, STRING, etc. Our analysis also includes alternative splicing and fusion Gene. Thebioinformatics analysis pipelineis as follows {{data.RenderAnchorLink("figure", "Figure", fn)}}.

{% endif %}

{% if data.ProjectInfo.ProjectType == 'DNBSEQ_RNAseq' or data.ProjectInfo.ProjectType == 'DNBSEQ_Strand-specific_RNAseq' %}

The data obtained by sequencing is called raw reads. Firstly,  reads with low quality, adapter contamination, and highly unknown N base are filtered. The filtered data is called clean reads. 

We will perform the RNA quantitation, the difference analysis, cluster analysis, time series analysis, enrichment analysis, etc. For mRNA, we will perform database annotations. The databases include NR, NT, GO, KOG, KEGG, UniProt, COG, Rfam, Pfam, TF, STRING, etc. Our analysis also includes alternative splicing and fusion Gene. Thebioinformatics analysis pipelineis as follows {{data.RenderAnchorLink("figure", "Figure", fn)}}.

{% endif %}

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/Analysis_flow_en.png
subTitle: Figure{{fn}} Information Analysis Flow
title: Analysis Flow
```
{% set fn = fn + 1%}

{% if data.FileExists("Filter") %}
## Data filtering

Affected by sample quality and species, the efficiency of ribosomes removal method may be unstable, and the contamination of ribosomes will affect the subsequent analysis, so we first use bowtie2 to map the clean reads to the ribosomal database, and remove the reads mapped to the ribosomal database, and use the retained data for subsequent analysis.

We define "dirty" raw reads as reads which contain adapters, high content of N bases and low quality reads. These reads need to be removed before data analysis. The filtering steps are as follows:

1) Remove reads with adapter;

2) Remove reads in which N bases are more than 1%;

3) Remove low-quality reads ((the percentage of low quality bases is over 40% in a read, we define the low quality base to be the base whose sequencing quality is no more than 20);

After filtering, the remaining reads are called "clean reads" and stored in FASTQ format used for downstream bioinformatics analysis.(see the FASTQ format description on the help page). The sequence file can be directly used for publication, public database submission, etc.

Software information:

SOAPnuke:

Version: v2.3

Parameters: -l 20 -q 0.4 -n 0.01

Official website: https://github.com/BGI-flexlab/SOAPnuke

{% endif %}

{% if data.FileExists("Assemble_denovo") %}

## Assemble

We use Trinity for assembly, and then use cdhit to de-redundant the assembled transcripts. Finally, we used the single-co
py direct homologous database BUSCO to evaluate the quality of the assembled transcripts. According to the assembly resul
ts, we detected the SSR of the transcript and designed primers.


Trinity:

Version: v2.13.2

Parameters: --seqType fq --max_memory 50G --group_pairs_distance 630  --no_version_check --full_cleanup  --min_contig_length 150  --CPU 8 --min_kmer_cov 3 --min_glue 3 --bfly_opts '-V 5 --edge-thr=0.1 --stderr '

Official website: https://github.com/trinityrnaseq/trinityrnaseq/


cdhit:

Version: 4.6.1

Parameters: -n 10 -c 0.95

Official website: https://sites.google.com/view/cd-hit

busco:

Version: v5.4.3

Parameters: -m tran -f --cpu 8 --offline

Official website: https://busco.ezlab.org/

{% endif %}


{% if data.FileExists("Assemble") %}

## Assembly

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/pic_Assembl.png
subTitle: Figure {{fn}} Assembly Flow
title: pic_Assembl
```
{% set fn = fn + 1%}

We use the map software HISAT2{{data.RenderReference(1)}} to align clean reads to the reference genome. HISAT2 is a fast and sensitive map software with higher accuracy. Reads mapped to the genome will be assembled with StringTie{{data.RenderReference(2)}}. For the assembly method, please refer to {{data.RenderAnchorLink("figure", "Figure", fn)}}{{data.RenderReference(19)}}. We use the chain-specific patterns of HISAT2 and StringTie for mapping and assembly, which can more accurately distinguish the reads from the positive strand or the negative strand, and more accurately count the number of transcripts and discover new genes.

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/pic_StringTie.png
title: lncRNA_pipe
subTitle: Figure {{fn}} Assembly strategy for StringTie
desc: a. Reads aligned to the reference genome. b. Construct a connection diagram of all possible transcripts in an aligned region. c, d. The overlap reads will be assembled into multiple transcripts based on the lines of the series connection diagram.
```
{% set fn = fn + 1%}


After transcript is reconstructed, we get all the transcript sequences in each sample. At this time, we use cuffcompare (one of the tools of Cufflinks{{data.RenderReference(20)}}) to compare these transcripts with known mRNA and lncRNA to obtain position relationship between each other.

For the assembly results of individual transcripts, we require each transcript to meet the following requirements:

1) FPKM >= 0.5

2) {{data.RenderHover("Coverage")}}> 1

3) Length> 200

By the lncRNA classification in the NONCODE{{data.RenderReference(22)}} database, we found in these five categories i,j,u,x,o{{data.RenderReference(21)}} also contain the most lncRNA , So we keep these transcripts.

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/pic_noncode.png
subTitle: Figure {{fn}} NONCODE lncRNA classification
title: pic_noncode
```
{% set fn = fn + 1%}

Genes expressed at low levels are not fully assembled in a single repeat due to insufficient sequencing depth, but a complete transcript can often be obtained after combining the results of multiple assembly. {{data.RenderReference(23)}} We use Cuffmerge (one tool of Cufflinks{{data.RenderReference(20)}}) to merge the results of assembly, and the merged transcripts are used as final results and used for subsequent analysis.

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/pic_merget.png
subTitle: Figure {{fn}} The image of merging transcripts
title: pic_merget
```

{% set fn = fn + 1%}

Software information:

HISAT2:

Version: v2.2.1

Parameters: --sensitive --no-discordant --no-mixed -I 1 -X 1000 -p 8 --rna-strandness RF

Official website: http://daehwankimlab.github.io/hisat2/

StringTie:

Version: v2.2.1

Parameters: -f 0.3 -j 3 -c 5 -g 100 -s 10000 -p 8

Official website: https://github.com/gpertea/stringtie

Cufflinks:

Version: v2.2.1

Parameters: Default parameters

Official website: http://cole-trapnell-lab.github.io/cufflinks

{% endif %}

{% if data.FileExists("Coding_Capacity") %}

## Coding Ability Prediction

After obtaining the new transcript, we will predict the coding ability of the transcript to distinguish between mRNA and lncRNA. We use three predictive softwares and two protein database.These three softwares are CPC{{data.RenderReference(3)}}, LGC and CNCI{{data.RenderReference(4)}}.The database are Pfam{{data.RenderReference(5)}} and SwissProt。

Pfam and SwissProt are protein databases. If the transcript can be aligned with the protein database, it is considered as mRNA, otherwise it is lncRNA. At least three of the five judgment methods are consistent and at least one protein database needs to be aligned ,then we can confirm that the transcript is mRNA.

The thresholds of the three software are as follows:

CPC2_threshold = 0,Transcripts greater than 0 are mRNA, and transcripts less than 0 are lncRNA.

CNCI_threshold = 0,Transcripts greater than 0 are mRNA, and transcripts less than 0 are lncRNA

LGC_threshold = 0,Transcripts greater than 0 are mRNA, and transcripts less than 0 are lncRNA

Software information:

CPC2:

Version: None

Parameters: Default parameters

Official website: http://cpc2.gao-lab.org/index.php

CNCI:

Version: None

Parameters: Default parameters

Official website: https://github.com/www-bioinfo-org/CNCI

LGC:

Version: LGC-1.0

Parameters: Default parameters

Official website: https://ngdc.cncb.ac.cn/lgc/calculator

{% endif %}

{% if data.FileExists("Quantification") %}

## Quantitative Analysis

We use Bowtie2{{data.RenderReference(6)}} to align clean reads to the reference sequence, and then use RSEM{{data.RenderReference(7)}} to calculate the expression of genes and transcripts. RSEM is a model generated by establishing reads, then distributes reads to different transcripts using maximum likelihood. We use a chain-specific model to distinguish the reads from the positive and negative strands, for more accurate quantification.

To compare the expression levels between samples, it is necessary to standardize the expression levels of genes. The standardized method used by RSEM is FPKM. The specific calculation formula of FPKM is as follows:

```image
url: {{data.ReportRoot}}/md/pic/pic_fpkm.png
title: pic_fpkm
```

Let FPKM(A) be the expression level of gene A, then C is the number of fragments uniquely aligned to gene A, N is the total number of fragments uniquely aligned to the reference gene, and L is the number of bases of gene A coding region. The FPKM method can eliminate the influence of the difference in gene length and sequencing amount on the calculated gene expression, and the calculated gene expression can be directly used to compare the differential expression between different samples.

Software information:

Bowtie2:

Version: 2.4.5

Parameters: -q --sensitive --dpad 0 --gbar 99999999 --mp 1,1 --np 1 --score-min L,0,-0.1 -I 1 -X 1000 --no-mixed --no-discordant  -p 8 -k 20 -N 1

Official website: http://bowtie-bio.sourceforge.net/bowtie2/index.shtml

RSEM:

Version: v1.3.1

Parameters: --forward-prob 0

Parameters: https://deweylab.github.io/RSEM/

{% endif %}

{% if data.FileExists("Coverage") %}

## Gene Coverage Analysis

We use Bowtie2{{data.RenderReference(6)}} to align clean reads to the reference gene, and then count the gene coverage.

Analysis steps:

1) Count the depth at each sites of the genes

2) Divide genes into 100 bins, and count the average coverage depth of each bin

3) Normalize the coverage depth of each bin (divide by the maximum of 100 bins)

4) Sliding in the gene area with 5 bin as the length and step length of 1; Calculate the average Coverage depth of these 5 bin,and finally find the variance of these averages.

5) The data drawn is the mean value, and these mean values are also normalized (divided by the maximum value).

{% endif %}

{% if data.FileExists("Difference/Group_Diff")%}

## Differences Between Groups

Differential analysis compare the expression value of  samples, excluding genes the expression value changes that not obvious, retaining genes with significant differences. Differences between groups compare the gene expression levels of two groups. A group can have multiple samples. Generally, one is the control group, and the other is the treatment group. We use the difference analysis software {{diffMethod}} for intergroup difference analysis.

Filtration conditions of gene with significant differences:

{% if diffMethod == 'DEseq2' %}
{{diffMethod}} : abs(log2FC) >= {{reportDataInfo.diff_cut_off_info.DEseq2.log2FC}} and Qvalue <= {{reportDataInfo.diff_cut_off_info.DEseq2.Qvalue}}
{% endif %}

{% if diffMethod == 'DEGseq' %}
{{diffMethod}} : abs(log2FC) >= {{reportDataInfo.diff_cut_off_info.DEGseq.log2FC}} and Qvalue <= {{reportDataInfo.diff_cut_off_info.DEGseq.Qvalue}}
{% endif %}

{% if diffMethod == 'NOIseq' %}
{{diffMethod}} : abs(log2FC) >= {{reportDataInfo.diff_cut_off_info.NOIseq.log2FC}} and Probability >= {{reportDataInfo.diff_cut_off_info.NOIseq.Probability}}
{% endif %}

{% endif %}

{% if data.FileExists("Difference/Sample_Diff")%}

## Differences Between Samples

Differential analysis compare the expression value of samples, excluding genes the expression value changes that not obvious, retaining gene with significant differences.  Differences between samples compare the gene expression value between two samples. Generally, one is a control sample and the another is a treatment sample. We use the difference analysis software PossionDis{{data.RenderReference(9)}} to analyze the difference between samples.

Filtration conditions of gene with significant differences:

PossionDis: abs(log2FC) >= 1 and {{data.RenderHover("FDR")}} <= 0.001

{% endif %}

{% if data.FileExists("Cluster_Pheatmap") %}

## Cluster Analysis

In order to find genes with similar expression trends, we use the software pheatmap to cluster the expression of genes. When clustering multiple samples, we take the union of expression genes.

Software information:

pheatmap:

Version: 1.0.12

Parameters: Default parameters

Official website: https://cran.r-project.org/web/packages/pheatmap/index.html

{% endif %}

{% if data.FileExists("Cluster_Mfuzz") %}

## Time Series Analysis

Clustering is a common method for analyzing gene expression data. There are two clustering algorithms: hard clustering and soft clustering. Hard clustering will cluster a gene into only one cluster, which is effective for gene clusters that are very separate. But for expression data, this method is not suitable because gene clusters often overlap. However, soft clustering can solve this problem. The time series analysis software Mfuzz{{data.RenderReference(26)}} used in this project is based on the soft clustering algorithm. Mfuzz is a software package based on R, which can cluster genes based on similar expression profiles, which helps to find genes with similar functions; at the same time, genes with the consistent expression trend may participate in the same biological process, which helps to find regulatory genes.

The membership of Mfuzz is a value of 0 to 1, which is used to evaluate whether a gene conforms to the change trend of the gene cluster. The closer its value is to 1, the more the gene trend conforms to the change trend of the gene cluster.

Software information:

Mfuzz:

Version: 2.54.0

Parameters: Default parameters

Official website: http://mfuzz.sysbiolab.eu

{% endif %}

{% if data.FileExists("Cluster_WGCNA") %}

## Co Expression Network Analysis - WGCNA

WGCNA (Weighted Gene Co expression Network Analysis) is a commonly used method for analyzing gene co expression networks. It can identify the association relationships between genes from large-scale gene expression data and aggregate genes with similar expression patterns into modules. Phenotypic association analysis is also one of the ultimate goals of WGCNA analysis, which can help us identify gene modules and core genes related to phenotypic characteristics. The analysis software is WGCNA.

Software information:

WGCNA:

Version: 1.72-1

Parameters: Default parameters

Official website: https://horvath.genetics.ucla.edu/html/CoexpressionNetwork/Rpackages/WGCNA/

{% endif %}

{% if data.FileExists("Annotation") %}

## Annotation

In order to better understand gene function, we will annotate the RNA. mRNA annotated databases include:NR,NT,GO,KOG,KEGG,uniprot,COG,Rfam,TF.lncRNA annotated databases include:miRBase,Rfam.

We used blast to align mRNA to NT; Use diamond to align mRNA to NR, KOG, KEGG, uniprot, COG, TF (animal transcription factor database is AnimalTFDB3, plant transcription factor database is PlantTFdb5); Obtain GO annotation information based on the align information of uniprot (including sprot and trembl). We used infernal to align lncRNA/mRNA to Rfam; Use diamond to align lncRNA to miRBase (miRNA precursors sequence).

Software information:

blast:

Version: 2.7.1

Parameters: -dust no -evalue 1e-15  -num_threads  5  -max_target_seqs 5 -outfmt "6 qseqid sseqid pident length mismatch gapopen qstart qend sstart send evalue bitscore stitle"

Official website: https://blast.ncbi.nlm.nih.gov/Blast.cgi

infernal:

Version: 1.1.4

Parameters: Default parameters

Official website: https://code.google.com/p/interproscan/wiki/Introduction

Diamond:

Version: v2.0.15.153

Parameters: Default parameters

Official website: https://github.com/bbuchfink/diamond

{% endif %}

{% if data.FileExists("Target")%}

## lncRNA Target Gene Analysis

The function of LncRNA is mainly achieved by cis or trans acting on the target gene.

Cis acting on the target gene prediction principle consider the function of lncRNA is related to the protein coding gene nearits coordinates,so mRNA adjacent to lncRNA are selected as target genes. And tran regulation is not dependent on the location relationship, we predict it by combining energy method.

If there is overlap between lncRNA and the target genes, we will classify the overlap in detail, which will help us understand the details of cis regulation. {{data.RenderReference(15)}}{{data.RenderReference(16)}}

Target gene analysis method:

Calculate the two correlation coefficients of lncRNA and mRNA, spearman and pearson, requiring spearman_cor>=0.6 and pearson_cor>=0.6.

Cis and trans analysis: lncRNA in the 10k upstream of 10k downstream of mRNA is determined as cis. Beyond this range, RNAplex will be used to analyze the binding energy of lncRNA and mRNA. If the binding energy is <-60, it is determined as tran.

{% endif %}

{% if data.FileExists("Rich") or data.FileExists("Rich_target") or data.FileExists("Rich_mfuzz") or data.FileExists("Rich_wgcna") %}

## Enrichment Analysis

Gene Ontology and {{data.RenderHover("KEGG")}} are the two most functional databases for gene classification. Therefore, we chose them to divide the gene functional modules, such as a {{data.RenderHover("GO")}} term as a functional module, or a pathway as a functional module.

In order to determine significant different genes and target genes of lncRNA on which functional modules are more focused, we will perform enrichment analysis.


If the candidate genes are significantly enriched in a functional module, it means that the proportion of candidate genes of the functional module is significantly higher than the background. The background is a reference, generally based on the overall as a background, such as the overall candidate gene ratio.


In order to quantitatively evaluate whether the candidate genes of a certain functional module are enriched, we used a hypergeometric model to calculate the enrichment of candidate genes for each functional module.

Each functional module corresponds to a p-value. The smaller the p-value, gene is more abundant on this functional module. Then we perform {{data.RenderHover("FDR")}} correction on p-value, we set {{data.RenderHover("FDR")}} <=0.01 functions as significant enrichment.

Filter conditions for significant enrichment: {{data.RenderHover("FDR")}} <=0.01

P-value calculation method:

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/pic_hyper.png
title: pic_hyper
```

Introduction to Hypergeometric Distribution: https://en.wikipedia.org/wiki/Hypergeometric_distribution

{% endif %}

{% if data.FileExists("GeneFusion") %}

## Fusion Gene

Gene fusion refers to the process by which all or part of the sequences of two genes are fused to each other to form a new gene. The mechanisms of forming a fusion gene is more complicated: including chromosome translocation, intermediate deletion, chromosome inversion and trans-splicing, etc, as shown in {{data.RenderAnchorLink("figure", "Figure", fn)}} . At present, gene fusion is mainly focused on cancer research, such as leukemia, prostate cancer, and breast cancer. Studies have shown that fusion genes are closely related to the occurrence and development of cancer, and the genes involved in the fusion are often some proto-oncogenes. The results of the fusion are diverse: a new fusion protein may be produced, which has the functions of the two parts of the fusion gene or has a new function; it is also possible that a proto-oncogene may be fused with a strong promoter of another gene cause the proto-oncogene to be high Expression, etc. Many research results have shown that fusion genes can be used as cancer-specific molecular signatures and potential drug targets, and have important significance in the clinical diagnosis and treatment of cancer.

{{data.RenderAnchor("figure", fn )}}

```image
url: {{data.ReportRoot}}/md/pic/Gene_Fusion_Types.png
subTitle: Figure{{fn}} Fusion gene type
desc: The reasons for the formation of fusion genes are: A) Chromosome translocation; B) Intermediate deletion; C) Chromo
some inversion.
```
{% set fn = fn + 1%}

Certain types of chromosomal aberrations and the resulting fusion genes are often used in cancer diagnosis to improve accuracy. Chromosome banding analysis, fluorescence in situ hybridization (FISH), reverse transcription polymerase chain reaction (RT-PCR) are commonly used in diagnostic laboratories, but due to the complexity of the cancer genome, these methods still have obvious shortcomings. At present, with the development of high-throughput sequencing technology, more accurate and effective methods have been brought to the detection of fusion genes.

We use Ericscript to analyse fusion gene.

Software information:

Version: v0.5.5

Parameters: Default parameters

Official website: https://github.com/databio/ericscript

{% endif %}

{% if data.FileExists("Alternative_Splicing") %}

## Alternative splicing

we use rMATS to detect differentially splicing gene (DSG) between samples. DSGs are regulated by alternative splicing (AS), which allows the production of a variety of different isofroms from one gene only. Changes in relative abundance of isoforms, regardless of the expression change, indicate a splicing-related mechanism. We detect five types of AS events, including Skipped Exon (SE), Alternative 5' Splicing Site (A5SS), Alternative 3' Splicing Site (A3SS), Mutually exclusive exons (MXE) and Retained Intron (RI). 

It is important to distinguish differential isoform relative abundance, from differential isoform expression. Changes in relative abundance of isoforms, regardless of the expression change, indicate a splicing-related mechanism. On the other hand, there can be measurable changes in the expression of isoforms across samples, without necessarily changing the relative abundance, which possibly indicates a transcription-related mechanism. We use rMATS to detect differentially splicing gene(that is differential isoform relative abundance between samples), a computational tool to detect differential alternative splicing events from RNA-Seq data, it calculates the inclusion isoform and skipping isofrom, shown as Figure. The statistical model of MATS calculates the P-value and false discovery rate (FDR) that the difference in the isoform ratio of a gene between two conditions, in our project, gene that with FDR <= 0.05 is defined as significant differentially splicing gene (DSG).

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/rMATS.png
title: rMATS
subTitle: Figure{{fn}} Calculation of relative abundance of various splicing types
```
{% set fn = fn + 1%}

Software information:

Version: v4.1.2

Parameters: Default parameters

Official website: https://rnaseq-mats.sourceforge.net/

{% endif %}

{% if data.FileExists("SnpIndel/SNP") or data.FileExists("SnpIndel/Indel") %}

## SNP and InDel Analysis

Based on the mapping results of the reference genome, we use GATK{{data.RenderReference(13)}} to detect the SNP {{data.RenderHover("SNP")}} and InDel {{data.RenderHover("InDel") }} for each sample. After filtering out low-quality site, the result is stored in VCF format. GATK is a software package developed by Broad Institute for next generation sequencing data analysis. It provides a variety of analytical methods, mainly focusing on mutation detection and genotyping. Robust architecture, powerful processing engine, high-performance computing and other characteristics make it widely used in various projects. The GATK analysis process is shown in {{data.RenderAnchorLink("figure", "Figure", fn)}}.

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/SNP_INDEL_workflow.png
title: SNP_INDEL_workflow
subTitle: Figure{{fn}} SNP and INDEL detection process by GATK
desc: Mainly divided into two steps: 1. Data pre-processing (alignment to genome-tag repeat sequence-segmentation for sequence-re-calibration of base. Note: The figure shows the alignment of reads to the genome using STAR software. But we use HISAT2 software in this program); 2. Mutation detection (HaplotypeCaller is used to detect SNP and Indel-filtering mutation site).
```
{% set fn = fn + 1%}

Software information:

GATK:

GATK: v4.2.6

Parameters (detection): Default parameters

Parameters (filter): -window 35 -cluster 3 --filter-name FS -filter "FS > 30.0" --filter-name QD -filter "QD < 2.0"

Official website: https://gatk.broadinstitute.org/hc/en-us

{% endif %}

{% if data.FileExists("cicrRNA")%}

## CircRNA Analysis

{{data.RenderAnchor("figure", fn)}}

```image
url: {{data.ReportRoot}}/md/pic/CIRI_pipeline.png
subTitle: Figure{{fn}} CIRI detect circRNA pipeine
title: CIRI_pipeline
```

This study uses CIRI{{data.RenderReference(27)}} software to predict CircRNA. CIRI is a software developed based on the BWA-MEM{{data.RenderReference(27)}} genome alignment algorithm,which which detects circRNA by two scans of the generated SAM (Sequence Alignment/Map) file. before and afterthe same read need to be cross-aligned to the genome to prove that they are junction reads at both ends of the circRNA. Therefore, during the first pass, CIRI will
filter the junction reads with paired chiastic clipping (PCC) signals and then limit the range of paired-end mapping (PEM). And GT-AG shear signal is further filtered for the junction reads. In the second scan of the SAM file, CIRI will detect new junction reads and eliminate false positive circRNAs caused by mismatches to homologous genes or repeats, thereby improving the sensitivity and accuracy of the software {{data.RenderReference (27)}}. For the detailed process, please refer to {{data.RenderAnchorLink("figure", "Figure", fn)}}.

{% set fn = fn + 1%}

Software information:

CIRI:

Version: v2.0.5

Parameters: Default

Official website: https://sourceforge.net/projects/ciri/

{% endif %}


## References
{{ data.RenderReferenceFoot() }}

{% endblock %}