Single cell code analysis - gynecological cancer single cell transcriptome and chromatin accessibility analysis 1

This article is 2021 Years published in MC Articles on , The article mainly talks about the use of single cell multiomics to analyze gynecological cancer .

Article title ：A multi-omic single-cell landscape of human gynecologic malignancies

doi：10.1016/j.molcel.2021.10.013

picture .png

The author provides a code link ：https://github.com/RegnerM2015/scENDO_scOVAR_2020

The author said that according to this link, you can roughly walk through the content ：ttps://github.com/RegnerM2015/scENDO_scOVAR_2020/wiki, Therefore, I also analyze the code basically according to this .

Introduction

This article is to analyze the single cell map of gynecological diseases . Therefore, the immediate treatment after surgical resection was selected 11 Tissue samples of individual ovarian and endometrial tumors , Dissociate single cells , Constructed a transcriptome with single cell resolution （SCRNA-SEQ） And chromatin accessibility (SCATAC-SEQ） Manual . This rare data set provides a solution to the complex cellular heterogeneity of these tumors , Enable researchers to link changes in chromatin accessibility to changes in gene expression . At the same time, these data provide information on how cancer cells reuse and acquire remote regulatory elements to drive the development of carcinogenic transcription patterns .

The following is the source of the sample ：

picture .png

The author also constructed the relevant experimental flow chart ：

picture .png

cellranger and cellranger atac processing

After dissociation of single cells , Get offline data , And then in cellranger On the analysis of （ Because the author chose 10X Single cell platform for labeling ）.

Transcriptome analysis mainly uses CellRanger (version 3.1.0).

## Add ： Construction of reference genome index 
cellranger mkref --genome= GRCh38-3.0.0 \
                 --fasta=GRCh38-3.0.0.fa \
                 --genes=GRCh38-3.0.0.gtf \
                 --ref-version=3.0.0

Copy

/Example_Patients1-2_scRNA-seq_mkfastq.sh

#!/usr/bin/env bash

#SBATCH --job-name HMGVCBGX9
#SBATCH -c 16
#SBATCH --mem 80g
#SBATCH --partition allnodes
#SBATCH --output HMGVCBGX9_demultiplex.job.out
#SBATCH --error HMGVCBGX9_demultiplex.job.err
## Data source file 
DATA=/datastore/nextgenout5/share/labs/bioinformatics/seqware/francolab_10x_copy
## Data result file 
OUT=/datastore/nextgenout5/share/labs/francolab/scRNA-seq_Endometrial.05.15.2019

## First integrate the offline files 
cellranger mkfastq --id=HMGVCBGX9 \
                   --run=${DATA}/190514_NS500270_0297_AHMGVCBGX9 \
                   --csv=${OUT}/Samplesheet.csv \
                   --qc \
                   --localcores=16

Copy

/Example_Patient1_scRNA-seq_CellRanger-count.sh

#!/usr/bin/env bash

##SBATCH: Submit the assignment 
#SBATCH --job-name 3533EL-RNA_F6
#SBATCH -c 16
#SBATCH --mem 80g
#SBATCH --partition allnodes
#SBATCH --output 3533EL-RNA_F6.cellranger-count.job.update.out 
#SBATCH --error 3533EL-RNA_F6.cellranger-count.job.update.err 

DATA=/datastore/nextgenout5/share/labs/francolab/Data

cellranger count  --id=3533EL-RNA_F6_Update \
                      --fastqs=./fastq_path/HMGVCBGX9/3533EL-RNA_F6 \
                      --transcriptome=${DATA}/refdata-cellranger-GRCh38-3.0.0 \ ## Reference genome 
                      --sample=3533EL-RNA_F6 \    ## According to sample Change the name of the content 
                      --localcores=16 \
                      --localmem=80

Copy

ATAC The analysis mainly uses CellRanger ATAC (version 1.2.0).

## Reference genome index construction 
## Build configuration files human.config
{
    organism: "human"        # The folder name can only be output after the species is clear 
    genome: ["GRCh38-3.0.0"]       
    input_fasta: ["/path/to/reference/GRCh38-3.0.0.fa"]
    input_gtf: ["/path/to/reference/GRCh38-3.0.0.gtf"]
    non_nuclear_contigs: ["chrM"] # Choose to do , Remove some annotation information from mitochondria 
    input_motifs: "/path/to/jaspar/motifs.pfm" #motif File path 
}
## structure ATAC Reference genome 
cellranger-atac mkref --config=/home/path/to/human.config

Copy

/Example_Patients1-2_scATAC-seq_mkfastq.sh

#!/usr/bin/env bash

#SBATCH --job-name H333JBGXB_2
#SBATCH -c 16
#SBATCH --mem 80g
#SBATCH --partition allnodes
#SBATCH --output H333JBGXB_demultiplex.2.job.out
#SBATCH --error H333JBGXB_demultiplex.2.job.err

## Data sources 
DATA=/datastore/nextgenout5/share/labs/bioinformatics/seqware/francolab_10x_copy
## Data result file 
OUT=/datastore/nextgenout5/share/labs/francolab/scATAC-seq_Endometrial.05.16.2019

cellranger-atac mkfastq --id=H333JBGXB_2 \
                        --run=${DATA}/190515_NS500270_0298_AH333JBGXB \
                        --csv=${OUT}/Samplesheet.2.csv \
                        --qc \
                        --localcores=16

Copy

/Example_Patient1_scATAC-seq_CellRanger-count.sh

#!/usr/bin/env bash

#SBATCH --job-name 3533EL-ATAC_A3
#SBATCH -c 16
#SBATCH --mem 80g
#SBATCH --partition allnodes
#SBATCH --output 3533EL-ATAC_A3.cellranger-count.job.update.out 
#SBATCH --error 3533EL-ATAC_A3.cellranger-count.job.update.err 

DATA=/datastore/nextgenout5/share/labs/francolab/Data

cellranger-atac count --id=3533EL-ATAC_A3_Update \
                      --fastqs=./fastq_path2/H333JBGXB/3533EL-ATAC_A3 \
                      --reference=${DATA}/refdata-cellranger-atac-GRCh38-1.2.0 \ ## Reference genome 
                      --sample=3533EL-ATAC_A3 \
                      --localcores=16

Copy

summary

It is mentioned in the author's important description of the data format provided used in this study ： Of each patient sample processed scATAC-seq The data is provided in the form of fragment files , or “ Be similar to bed Form file for , Each line represents the unique ATAC-seq fragment ”. More information about this file format , Please visit Cell Ranger Website :https://support.10xgenomics.com/single-cell-atac/software/pipelines/latest/output/fragments. We provide fragment files , instead of cellranger-atac Generated filtered peak barcode matrix , Because we use fragment files as in ArchR R Executed in package scATAC-seq The starting input of the analysis (Granja wait forsomeone ,2021 year ). We didn't use by cellranger-atac Generated filtered peak barcode matrix , Because the algorithm calls the peak in a pseudo batch way ( That is, use all signals from all cells in the sample ). This pseudo volume method effectively masks cell type specific patterns in chromatin accessibility , And will damage rare cell types ATAC Signal contribution (Granja wait forsomeone ,2021).

This problem mentioned by the author also encountered this situation in my analysis , At the same time, we also use this method for analysis , It is also more conducive to subsequent analysis .

So when analyzing single cell data , The first is to analyze according to the basic process , Then judge the data quality , Then on cellranger Change the parameters of the software , It is different from the previous software , Therefore, a large number of parameters need to be adjusted in the first step .