hwanglab / apa_atingLab2019

DNA methylation regulates alternative polyadenylation via CTCF and the cohesin complex (2019 paper submission)

Download APA reproducible package

mkdir -p $HOME/projects
cd $HOME/projects
git clone https://github.com/hwanglab/apa_2019.git

• In this guideline, the package base directory is "$HOME/projects/apa_atingLab2019". In the following steps, it is necessary to modify scripts to be matched with your local setup.

Prerequisite

1. python 2.7

   cd apa_atingLab2019
   pip install --user -r python_requirements.txt #Alternatively, use conda environment to install packages listed in the requirements.txt
   

2. R 3.4.3 packages

   handy (refer to https://github.com/jeffbhasin/handy for installation)
   goldmine
   argparse
   BSgenome.Hsapiens.UCSC.hg19
   corrplot
   data.table
   DEXSeq
   e1071
   edgeR
   ggbio
   ggplot2
   gridExtra
   IlluminaHumanMethylation450kanno.ilmn12.hg19
   matrixStats
   openxlsx
   parallel
   reshape
   RPMM
   Rsamtools
   Rsubread
   stringr
   ggseqlogo
   JASPAR2018
   TFBSTools
   UpSetR
   VennDiagram
   wateRmelon
   

3. programs (the binary files should be in $PATH)

   bowtie2
   mosdepth
   parallel
   picard
   samtools
   sra-tools
   STAR
   twoBitToFa
   wget
   wigToBigWig
   xz
   

4. add R library path to your bash shell environment file ($HOME/.bashrc or $HOME/.bash_profile),

   export R_UTIL_APA=$HOME/projects/apa_atingLab2019/resource/libs
   export PADD_GIT=$HOME/projects/apa_atingLab2019/01_polyAseq/paddle-git
   

PolyA-seq data processing

1. Configure program paths and resource files

   bash ./01_resource_prep.sh

2. Download PolyA-seq FASTQ files. Note that, as of 05/21/2019, the SRAs,

   • SRP083252 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE86178)
   • SRP083254 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE86180)

   is not public. Secure tokens are available only reviewers to access processed data. Any raw data is not available yet until it becomes public.

   cd 01_polyAseq
   bash ./02_get_fastq.sh
   

3. Open the polyA-seq pipeline configuration file (./configs/program.conf) and edit the following two section aligned with your local setup,

   [bowtie2-build]
   bin_path = bowtie2-build
   ref_hg19_path = ~/projects/apa_atingLab2019/resource/ref/hg19_PhiX.fa
   ref_hg19_prefix = ~/projects/apa_atingLab2019/resource/ref/hg19_PhiX
   chrom_size = ~/projects/apa_atingLab2019/resource/ref/ChromSizes.hg19.txt
   
   [ensembl]
   biotype_table = ~/projects/apa_atingLab2019/resource/ensembl/ensembl_gene_biotypes.csv
   gtf_file = ~/projects/apa_atingLab2019/resource/ensembl/Homo_sapiens.GRCh37.87.gtf

4. Open the script (03_polyaseq.sh) and modify the package base directory

   export PPD=$HOME/projects/apa_atingLab2019/01_polyAseq
   

5. Running polyA-seq processing pipeline

   bash ./03_polyaseq.sh
   

6. Transcription factor binding analysis in the genomic regions in between polyA sites at each 546 APA gene

   Rscript ./04_enrich_perms_100k_select.r
   

7. Plot sequence logos for the highly enriched TFs

   Rscript ./05_plot_seqlogos.r
   

RNA-seq data processing

1. Contact authors to download RNA-seq FASTQ files.

   cd apa_atingLab2019/02_mRNAseq
   bash ./01_get_fastq.sh
   

2. Run STAR

   bash ./02_run_star.sh
   

3. Collect a basic alignment statistics

   Rscript ./03_get_alignment_stats.r
   

4. Analyzing the APA regulatory gene expressions. If the number of CPU's available is 8,

   bash ./04_apafactorexp.sh 8
   

ChIP-seq data processing

1. We use ENCODE TF and Histone ChIP-Seq processing pipeine from Kundaje lab (e.g., chipseq.bds.20180726_175025_830). If you don't have the pipeline, then,

   1. Download the pipeline at https://github.com/kundajelab/chipseq_pipeline

   2. Follow the installation instruction. The installation directories used in this guideline are,

      • $HOME/apps/chipseq/chip-seq-pipeline2 # ENCODE pipeline path
      • $HOME/.bds # bds path
      • $HOME/apps/miniconda3/bin #miniconda3 path
      • /opt/bds_pipeline_genome_data/aquas_chipseq_species.conf #species configuration file
      • Make sure that the installation is successful

   3. Come back to the APA project directory, change to ChIP-seq working directory, and download ChIP/MBD-seq FASTQ files.

   • https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE131606
   • https://www.ncbi.nlm.nih.gov/sra/SRP001414

   As of 05/21/2019, note that ChIP-seq raw fastq files are not in public. cd $HOME/projects/apa_atingLab2019/03_chipseq bash ./00_get_fastq.sh

   1. Open the bash script (chipseq_pipeline_arg.sh) to modify the paths matched with your local setup.

      progd="$HOME/apps/chipseqs/TF_chipseq_pipeline"
      species_conf="/opt/bds_pipeline_genome_data/aquas_chipseq_species.conf"
      
      export PATH=$HOME/.local/bin
      export PATH=$HOME/.bds:$PATH
      export PATH=$HOME/apps/miniconda3/bin:$PATH
      export PATH=/usr/lib64/ccache:/usr/local/bin:/usr/local/sbin:/usr/bin:/usr/sbin:$PATH
      

2. Launch the ENCODE ChIP-seq pipeline. Note that it will take for a while to complete the job.

   bash ./01_chipseq_by_encode.sh 2>&1 | tee logs/01_chipseq_by_encode_$(date '+%Y-%m-%d-%H').log
   

3. Process MBD-seq FASTQ files.

   bash ./02_mbdseq_by_bt2.sh 2>&1 | tee logs/02_mbdseq_by_bt2_$(date '+%Y-%m-%d-%H').log
   

4. Generate bigWig files for ChIP-seq alignment files for visualization. If you have a ftp server to host bigWig files to import to UCSC genomebrowser, open the script and modify ftp://account:passwd@hostname/apa_atingLab2019/chipseq_bigwigs.

   bash ./03_chipseq_tags_to_bigwig.sh
   bash ./04_input_tags_to_bigwig.sh
   

5. Collect some basic alignment statistics

   bash ./05_get_align_stats.sh
   

6. Perform a differential ChIP-seq binding site analysis using MANorm.

   bash ./06_manorm_diff_chipseq.sh
   Rscript ./06a_merge_manorm_report.r
   Rscript ./06b_store_manorm_sites.r
   

7. Visualization with ChIP-seq differential binding sites. Note that b03_nmf.sh requires that MATLAB is installed and should be in $PATH.

   Rscript ./b01_get_apa_intby_chipseq.r
   Rscript ./b02_uniq_xcvg_reg.r
   bash ./b03_nmf.sh
   Rscript ./b04_gen_heatmap_filtw.r
   Rscript ./b05_cluster_analysis.r
   

TCGA

TCGA data access permission is required to complete this analysis. Consider the following scripts to figure out which procedures/parameters were used in the paper for reference. Contact us for more help.

1. Change to TCGA working directory and download a preprocessed data (TCGA methylation beta values and pA usage ratio on APA genes of interest).

   cd $HOME/projects/apa_atingLab2019/04_tcga
   bash ./01_resource_prep.sh
   less -S 546goi_all/github_Supplementary_Table_S6.tsv #table
   

2. Access TCGA RNA-Seq BAM files and 4500 Infinium methylation array. Compute predicted polyA usage and normalized methylation level (beta value). Visualize the correlation with adjusted P-value in scatter plots and gene track along with ChIP-seq binding site read pileup from the cell line model designed.

   ./02a_tcga_analy_apa_paur.r
   ./02b_tcga_analy_apa_paur_plot.sh
   

3. To summarize/visualize the correlation between polyA usage predicted from mRNA-seq and methylation level,

   cd ./03_tcga_summary
   Rscript ./01_collapse_corr_mpts_max_corr_per_cohort.r
   Rscrtip ./02_summary_corr_table.r
   

4. To generate a box plot of polyA usage and methylation level in HEATR2 per tumor stage.

   cd ../04_tcga_clinical_info
   bash ./runme.sh
   

Reference