Methylation Analysis - C.virginica Gonad MBD with Varying Read Subsets with Bismark on Mox

2019
Miscellaneous
Author

Sam White

Published

March 13, 2019

Steven asked for the following analysis in this GitHub Issue using Yaamini’s C.virginica MBD samples:

  • Average read count for a single library
  • 50% of average read count for a single library
  • All data
  • 50% of all data

Questions to addressed Mean CpG Coverage % CpG Coverage at 5x % CpG Coverage at 10x Any “Report” information automatically calculated by Bismark.

I made a bisulfite converted genome using the NCBI GCF_002022765.2_C_virginica-3.0 version.

Set up the following SBATCH script to run on Mox:

NOTE: It seems as though the if/else statement didn’t work properly and resulted in a duplicated analysis of the “half total reads”. The “total reads” will be run independently.

2nd NOTE: Actually, the “problem” was that I was calculating read counts based on combining R1 and R2 reads. However, Bismark uses the subsetting option to run on read pairs (i.e. -u 1000 would run 1000 read pairs which is 2000 reads). So, my usage was incorrect. Fixed and re-ran.

3rd NOTE: I generated cytosine coverage files after the fact on a Mox build node and it is not part of the SBATCH script. I ran the following command in each of the Bismark output folders:

/gscratch/srlab/programs/Bismark-0.21.0_dev/coverage2cytosine \
cvir_bsseq_all_pe_R1_bismark_bt2_pe.bismark.cov.gz  \
--output bismark_cytosine_coverage.txt \
--genome_folder /gscratch/srlab/sam/data/C_virginica/genomes/GCF_002022765.2_C_virginica-3.0

#!/bin/bash
## Job Name
#SBATCH --job-name=bismark
## Allocation Definition
#SBATCH --account=coenv
#SBATCH --partition=coenv
## Resources
## Nodes
#SBATCH --nodes=1
## Walltime (days-hours:minutes:seconds format)
#SBATCH --time=10-00:00:00
## Memory per node
#SBATCH --mem=120G
##turn on e-mail notification
#SBATCH --mail-type=ALL
#SBATCH --mail-user=samwhite@uw.edu
## Specify the working directory for this job
#SBATCH --workdir=/gscratch/scrubbed/samwhite/outputs/20190312_cvir_gonad_bismark

# Load Python Mox module for Python module availability

module load intel-python3_2017

# Document programs in PATH (primarily for program version ID)

date >> system_path.log
echo "" >> system_path.log
echo "System PATH for $SLURM_JOB_ID" >> system_path.log
echo "" >> system_path.log
printf "%0.s-" {1..10} >> system_path.log
echo ${PATH} | tr : \\n >> system_path.log


# Directories and programs
wd=$(pwd)
bismark_dir="/gscratch/srlab/programs/Bismark-0.19.0"
bowtie2_dir="/gscratch/srlab/programs/bowtie2-2.3.4.1-linux-x86_64/"
samtools="/gscratch/srlab/programs/samtools-1.9/samtools"
reads_dir="/gscratch/srlab/sam/data/C_virginica/bsseq"

## genomes

genome="/gscratch/srlab/sam/data/C_virginica/genomes/GCF_002022765.2_C_virginica-3.0/"

## Concatenated FastQ Files
R1="${wd}/cvir_bsseq_all_pe_R1.fastq.gz"
R2="${wd}/cvir_bsseq_all_pe_R2.fastq.gz"

## FastQ files lists
### FastQ files were previously trimmed in preparation for use with Bismark:
### https://robertslab.github.io/sams-notebook/posts/2018/2018-04-11-trimgalorefastqcmultiqc-trim-10bp-53-ends-c-virginica-mbd-bs-seq-fastq-data/
R1_list="${wd}/cvir_bsseq_pe_all_R1.list"
R2_list="${wd}/cvir_bsseq_pe_all_R2.list"

# Initialize variables
total_reads=0
avg_reads=0
half_avg_reads=0
half_total_reads=0
reads_set_array=()
R1_array=()
R2_array=()


## Save FastQ files to arrays
R1_array=(${reads_dir}/*_R1_*.fq.gz)
R2_array=(${reads_dir}/*_R2_*.fq.gz)

# Number of libraries by counting elements in array
num_libs=$(echo ${#R1_array[@]})

# Add names to array
## Two IMPORTANT notes:
## 1. The "total reads" must be the last element of the array.
## 2. The order of this array must match the "reads_set_array" below!
reads_set_names_array=("avg_reads" "half_avg_reads" "half_total_reads" "total_reads")

# Check for existence of previous concatenation
# If they exist, delete them

for file in ${R1} ${R1_list} ${R2} ${R2_list} library_counts.txt
do
  if [ -e ${file} ]; then
    rm ${file}
  fi
done


printf "%s\t%s\n\n" "LIBRARY" "COUNT" >> library_counts.txt

# Determine total reads counts from all libraries
## Iterates through arrays and determines read counts
## by counting lines in FastQ and dividing by 4.
## Each loop adds the read1 read counts to the total
## Only uses R1 because Bismark interprets subsetting value as read pairs
for fastq in "${!R1_array[@]}"
do
  lib_count=0
  R1_fastq=${R1_array[fastq]}
  lib_name=$(echo ${R1_fastq} | awk -F'_' '{ print $3 }')
  R1_count=$(echo $(zcat ${R1_fastq} | wc -l)/4 | bc)
  lib_count=$(echo ${R1_count})
  printf "%s%s\t%s\n" "library_" "${lib_name}" "${lib_count}" >> library_counts.txt
  total_reads=$(echo ${R1_count} + ${total_reads}| bc)
done

# Calcuations for different read amounts desired for analysis.
avg_reads=$(echo ${total_reads}/${num_libs} | bc)
half_avg_reads=$(echo "${avg_reads} / 2" | bc)
half_total_reads=$(echo "${total_reads} / 2" | bc)


# Store calculated values in array
## IMPORTANT note:
## The order of this array must match "reads_set_names_array" above!
reads_set_array=(${avg_reads} ${half_avg_reads} ${half_total_reads} ${total_reads})

# Loop to record counts/calculations
for name in ${!reads_set_names_array[@]}
do
  count_name=${reads_set_names_array[name]}
  counts=${reads_set_array[name]}
  printf "%s\t%s\n" "${count_name}" "${counts}" >> library_counts.txt
done

# Concatenate R1 reads and generate lists of FastQs
for fastq in ${reads_dir}/*R1*.gz
do
  echo ${fastq} >> ${R1_list}
  cat ${fastq} >> ${R1}
done

# Concatenate R2 reads and generate lists of FastQs
for fastq in ${reads_dir}/*R2*.gz
do
  echo ${fastq} >> ${R2_list}
  cat ${fastq} >> ${R2}
done

# Run bismark using bisulftie-converted genome
# Converted genome from 20190222 by me:
# https://robertslab.github.io/sams-notebook/posts/2019/2019-02-21-Data-Management---Create-C.virginica-Bisulfite-Genome-wit-Bismark-on-Mox/
## Loops through the various read sets
## performs each analysis in respective subdirectory
## When loop encounters last element of the array, then Bismark is run without the
## -u subsetting option.
for set in "${!reads_set_names_array[@]}"
do
  set_name=${reads_set_names_array[set]}
  reads_set=${reads_set_array[set]}
  mkdir ${set_name}_bismark
  cd ${set_name}_bismark
  if [ ${set} -eq $(( ${#reads_set_names_array[@]} - 1 )) ]; then
    ${bismark_dir}/bismark \
    --path_to_bowtie ${bowtie2_dir} \
    --genome ${genome} \
    --non_directional \
    --score_min L,0,-0.6 \
    -p 28 \
    -1 ${R1} \
    -2 ${R2} \
    2> ${set_name}_summary.txt
  else
    ${bismark_dir}/bismark \
    --path_to_bowtie ${bowtie2_dir} \
    --genome ${genome} \
    --non_directional \
    --score_min L,0,-0.6 \
    -u ${reads_set} \
    -p 28 \
    -1 ${R1} \
    -2 ${R2} \
    2> ${set_name}_summary.txt
  fi
  # Methylation extraction

  ${bismark_dir}/bismark_methylation_extractor \
  --bedgraph \
  --counts \
  --scaffolds \
  --remove_spaces \
  --multicore 28 \
  --buffer_size 75% \
  *.bam

  # Bismark processing report

  ${bismark_dir}/bismark2report

  #Bismark summary report

  ${bismark_dir}/bismark2summary

  # Sort files for methylkit and IGV

  find *.bam \
  | xargs basename -s .bam \
  | xargs -I{} ${samtools} \
  sort \
  --threads 28 \
  {}.bam \
  -o {}.sorted.bam

  # Index sorted files for IGV
  # The "-@ 28" below specifies number of CPU threads to use.

  find *.sorted.bam \
  | xargs basename -s .sorted.bam \
  | xargs -I{} ${samtools} \
  index -@ 28 \
  {}.sorted.bam
  cd ${wd}
done

RESULTS

This took a pretty long time to run (6days 16hrs):

6 days, 16hrs runtime

Additionally, I don’t have the time at the moment to actually finish the analysis to evaluate differences in coverage. I’m posting notebook entry to make data available (belatedly) and will get some sort of comparison analysis done in the near future…

Output folder:

Summary reports for each of the configurations (HTML):