Sage-Bionetworks/amp-rnaseq

Name: amp-rnaseq

Owner: Sage Bionetworks

Description: Code for reprocessing the RNAseq data from AMP-AD.

Forked from: jaeddy/amp-rnaseq

Created: 2017-05-04 19:37:33.0

Updated: 2017-05-26 22:36:37.0

Pushed: 2017-05-04 19:41:53.0

Homepage: null

Size: 52

Language: Jupyter Notebook

GitHub Committers

User	Most Recent Commit	# Commits

Other Committers

User	Email	Most Recent Commit	# Commits

README

amp-rnaseq

Code for reprocessing the RNAseq data from AMP-AD.

All of the code in this repository includes strong dependencies on the software and filesystem organization within the MSSM Minerva scientific computing environment. Example LSF or shell commands are provided to illustrate how we used the scripts to reprocess the AMP-AD RNA-seq data.

Retrieval of site-submitted BAMs from Synapse

Code not provided for this step, as we used basic synapse get commands for all relevant Synapse IDs in each project (ROSMAP, MSSM, Mayo TCX, Mayo CBE).

Re-aligning & counting reads with STAR

Each of these scripts converts aligned reads from the downloaded Synapse BAM file into unmapped FASTQ format (for MSSM, unmapped reads not included in the BAM file are added from a separate FASTQ file) and re-aligns the reads using STAR while also counting reads per gene.

Running the scripts below produces a folder for each sample containing outputs from STAR, most importantly (1) aligned reads in a file labelled *Aligned.out.bam; and (2) gene counts in a file labelled *ReadsPerGene.out.tab.

Tool versions

• STAR: 2.5.1b
• Picard: 2.2.4

Usage

Options used for LSF bsub commands may vary depending on available system resources and file organization.

run_star_rosmap.sh: takes only one input, the BAM file obtained from Synapse for the current sample

 \
-q alloc -P acc_AMP_AD -J star -W 8:00 \
-n 4 -R rusage[mem=9000] -R span[ptile=4] \
-o %J.stdout -e %J.stderr \
-L /bin/bash <path-to-code-folder>/amp-rnaseq/run_star_rosmap.sh \
    406_120503.bam

run_star_mssm.sh: takes two inputs for each sample — (1) the BAM file obtained from Synapse; and (2) the FASTQ file (also from Synapse) containing unmapped reads

 -q alloc -P acc_AMP_AD -J star -W 8:00 \
-n 4 -R rusage[mem=9000] -R span[ptile=4] \
-o %J.stdout -e %J.stderr \
-L /bin/bash <path-to-code-folder>/run_star_sinai.sh \
    BM_10_546.accepted_hits.sort.coord.bam \
    <path-to-unmapped-fastqs-folder>/BM_10_546.unmapped.fq.gz

run_star_mayo.sh: takes two inputs for each sample — (1) the BAM file obtained from Synapse; and (2) the abbreviated brain region (TCX or CBE)

 -q alloc -P acc_AMP_AD -J star -W 8:00 \
-n 4 -R rusage[mem=9000] -R span[ptile=4] \
-o %J.stdout -e %J.stderr \
-L /bin/bash <path-to-code-folder>/amp-rnaseq/run_star_mayo.sh \
    6976_TCX.snap.bam \
    TCX

Combining gene count data

The combine_counts_study.R combines gene count profiles for all individual samples into a single matrix (rows of features by columns of samples). The example commands all assume that the current directory contains all read count files (ending in ReadsPerGene.out.tab) from STAR, enumerated using the find command.

The script outputs a tab-delimited matrix file labeled as <out_prefix>_all_counts_matrix.txt.

ROSMAP:

h-to-code-folder>/amp-rnaseq/bin/combine_counts_study.R \
--out_prefix ROSMAP \
$(find . -name "*ReadsPerGene*" -type f)

MSSM: (non-default sample suffix to strip trailing characters from BAM

file name and keep only sample/specimen ID)

h-to-code-folder>/amp-rnaseq/bin/combine_counts_study.R \
--out_prefix MSSM \
--sample_suffix "\.accepted_hits.*" \
$(find . -name "*ReadsPerGene*" -type f)

Mayo: (use “Mayo_CBE” as output prefix for CBE samples)

h-to-code-folder>/amp-rnaseq/bin/combine_counts_study.R \
--out_prefix Mayo_TCX \
$(find . -name "*ReadsPerGene*" -type f)

Generating alignment metrics with Picard

Similar to the run_star_* scripts above, the run_picard.sh was used to generate alignment metrics for an individual sample based on the output BAM file from STAR. Specifically, the script computes metrics with the CollectAlignmentSummaryMetrics and CollectRnaSeqMetrics Picard modules, then combines metrics into a single table for the current sample. The same script can be used for all studies.

Running the script produces a folder for each sample containing outputs from Picard as well as the combined table (generated using the script combine_metrics_sample.py in the bin/ folder), the latter of which is labelled as *_picard.CombinedMetrics.csv.

Tool versions:

• Picard: 2.2.4

Usage:

The commands below all assume that the current directory is a folder with subfolders for each sample, each containing the output BAM file from STAR.

ROSMAP:

 -q alloc -P acc_AMP_AD -J picard -W 8:00 \
-n 4 -R rusage[mem=8000] -R span[ptile=4] \
-o %J.stdout -e %J.stderr \
-L /bin/bash <path-to-code-folder>/code/amp-rnaseq/run_picard.sh \
    406_120503/406_120503Aligned.out.bam

MSSM:

 -q alloc -P acc_AMP_AD -J picard -W 8:00 \
-n 4 -R rusage[mem=8000] -R span[ptile=4] \
-o %J.stdout -e %J.stderr \
-L /bin/bash <path-to-code-folder>/amp-rnaseq/run_picard.sh \
    BM_10_546.accepted_hits.sort.coord/BM_10_546.accepted_hits.sort.coordAligned.out.bam

Mayo: (example provided for TCX, but usage is the same for CBE)

 -q alloc -P acc_AMP_AD -J picard -W 8:00 \
-n 4 -R rusage[mem=8000] -R span[ptile=4] \
-o %J.stdout -e %J.stderr \
-L /bin/bash <path-to-code-folder>/amp-rnaseq/run_picard.sh \
    6976_TCX/6976_TCXAligned.out.bam

Combining alignment metrics data

The combine_metrics_study.R combines metrics tables for all individual samples into a single matrix (rows of samples by columns of metrics) per study. The example commands all assume that the current directory contains all metrics count files (ending in CombinedMetrics.csv) from the run_picard script, enumerated using the find command.

The script outputs a tab-delimited matrix file labeled as <out_prefix>_all_metrics_matrix.txt.

ROSMAP:

h-to-code-folder>/amp-rnaseq/bin/combine_metrics_study.R \
--out_prefix ROSMAP \
$(find . -name "*CombinedMetrics*" -type f)

MSSM:

h-to-code-folder>/amp-rnaseq/bin/combine_metrics_study.R \
--out_prefix MSSM \
$(find . -name "*CombinedMetrics*" -type f)

Mayo: (use “Mayo_CBE” as output prefix for CBE samples)

h-to-code-folder>/amp-rnaseq/bin/combine_metrics_study.R \
--out_prefix Mayo_TCX \
$(find . -name "*CombinedMetrics*" -type f)

---

The following processing steps are in progress and will be described in more detail in a subsequent data/code release.

Quantifying transcript abundance with sailfish

run_sailfish_rosmap.sh run_sailfish_mssm.sh run_sailfish_mayo.sh

Converting bootstrap file formats

run_convertboostraps.sh

Computing quantification variance from bootstraps

run_calcsailfishvar.sh

Combining transcript quantification data

combine_quant_study.R