RNAseq pipeline

From ZhangLabWiki
Jump to navigation Jump to search

TSCC pipeline for single-cell RNAseq analysis[edit]

  • The input data files should follow the SOP of UCSD SCAP center. Basically, all PE gzip compressed fastq files for one cell are stored in one folder (the name starts with "Sample_"). One folder per cell or sequencing library.
  • We are currently using hg19 for human and mm10 for mouse. LifeTech's ERCC transcripts were included. For transcript models, we are using GENCODE V14 for human, and UCSC Genes for mouse.

Data transfer[edit]

  • Single-cell RNAseq data are uploaded by Illumina to the Tabit server (tabit.ucsd.edu). Note that this is a storage server, and only accessible through sftp.
  sftp Illumina@tabit.ucsd.edu
  # Type in password
  # The data files are typically in RNASeq or SampleGroup directory
  mget -r TARGET_FILE_OR_FOLDER
  • Mapped data in TSCC should be transferred to genome-miner, which can be done by scp.
  scp -r TARGET_FILE_OR_FOLDER YOUR_USER_NAME@genome-miner.ucsd.edu:FULL_NAME_OF_THE_FOLDER

Smart-seq pipeline[edit]

  • This pipeline is designed for mapping and transcript quantification of Smart-seq data.
  • It assumes that the data cover the full length of the transcripts, and have no strand specificity.

Pipeline A.[edit]

  • This pipeline uses Tophat2 and Cufflink2.
  • We use one script to submit jobs for all samples in one folder, one job per sample. Here is how to start the job.
    # First, copy the job submission script to your own path, such as your HOME/bin
   cp /home/k4zhang/bin/submit_tophat_jobs.pl MY_HOME/bin
    # Next edit this script, replace the email address "kun.zhang.ucsd@gmail.com" with your own address. 
    # This only need to be done once. 
    # As a rehearsal, you can copy two test data sets to your own folder.
   cp -r /home/k4zhang/my_oasis_tscc/test_data MY_FOLDER
    # Now you can change to this folder.
   cd MY_FOLDER
    # Then you can submit all the jobs to process the samples in this folder. 
    # For human data, replace mm10 with hg19.
   submit_tophat_jobs.pl mm10
    # To check the status, use qstat. 
    # You should also receive emails when the jobs are started or completed.
   qstat | grep YOUR_USER_NAME
    # The results are in tophat_xxx folder (one folder per sample).
    # After tophat mapping and cufflink quantification, the downstream analyses
    #   are typically done on genome-miner, because it is more interactive and 
    #   less time consuming. For this, use sftp or scp to transfer all 
    #   tophat_xxx folders to genome-miner
    # To quickly get an overview of mapping stats and number of transcripts called,
    #    run the following script within the GROUP_DATA_DIR that contains all tophat_xxx folders.
   /home/kunzhang/RNAseq/SCAP/get_mapping_stats.pl GROUP_DATA_DIR > group_name_mapping_stats.txt

Pipeline B (~10x faster, but requires more RAM, doesn't work on genome-miner)[edit]

  • This pipeline uses STAR and Cufflink2.
  • We use one script to submit jobs for all samples in one folder, one job per sample. Here is how to start the job.
    # First, copy the job submission script to your own path, such as your HOME/bin
   cp /home/k4zhang/bin/batch_STAR_cufflink2.pl MY_HOME/bin
    # Next edit this script, replace the email address "kun.zhang.ucsd@gmail.com" with your own address. 
    # This only need to be done once. 
    # As a rehearsal, you can copy two test data sets to your own folder.
   cp -r /home/k4zhang/my_oasis_tscc/test_data MY_FOLDER
    # Now you can change to this folder.
   cd MY_FOLDER
    # Then you can submit all the jobs to process the samples in this folder. 
    # For human data, replace mm10 with hg19.
   batch_STAR_cufflink2.pl mm10
    # To check the status, use qstat. 
    # You should also receive emails when the jobs are started or completed.
   qstat | grep YOUR_USER_NAME
    # The results are in xxx_mapped folder (one folder per sample).
    # To quickly get an overview of mapping stats and number of transcripts called,
    #  transfer all xxx_mapped folders to genome-miner in a folder (GROUP_DATA_DIR)
    #    run the following script  
   /home/kunzhang/RNAseq/SCAP/scripts/get_STAR_mapping_stats.pl GROUP_DATA_DIR > group_name_mapping_stats.txt
Pipeline B Version 2[edit]
  • Simply replace /home/k4zhang/bin/batch_STAR_cufflink2.pl with /home/k4zhang/bin/batch_STAR_cufflink2_HTseq.pl
  • What's new?
    • It handles both PE and SE reads.
    • It uses HTSeq to report raw read counts in both exonic and intronic regions, which is more appropriate for human nuclei data.
  • Three scripts for generating summary reports on genome-miner: