Skip to content

Latest commit

 

History

History
51 lines (25 loc) · 2.54 KB

analysis_workflow.md

File metadata and controls

51 lines (25 loc) · 2.54 KB

Analysis Workflow

Warning: This file is still under construction. I promise it will be completed before this study is published.

First you'll need to clone this directory. The following workflow is designed to be run from the root directory of the repository.

git clone https://github.com/SchlossLab/Baxter_glne007Modeling_GenomeMed_2015.git

Acquiring the raw data

Download .sra files from the NCBI Sequence Read Archive. The accession number for this study is SRP062005.

wget -r -np -nd -k -P data/ ftp://ftp-trace.ncbi.nih.gov/sra/sra-instant/reads/ByStudy/sra/SRP/SRP062/SRP062005/

Extract fastq files from the sra files. You will need to have the SRA Toolkit installed on your system. This command will also delete the sra files so that they don't waste storage space.

for FILE in data/*.sra; do fastq-dump --split-files --outdir data/ $FILE && rm -f $FILE; done

Sequence Curation

The next step is sequence curation with mothur.

Download and unzip reference files.

wget -P data/ http://www.mothur.org/w/images/b/be/Silva.nr_v123.tgz
tar zxv -C data/ -f data/Silva.nr_v123.tgz
wget -P data/ http://mothur.org/w/images/2/24/Trainset10_082014.pds.tgz
tar zxv -C data/ -f data/Trainset10_082014.pds.tgz

The data/ directory should contain a file called glne007.files. This file tells mothur which fastq files belong to each sample. You may notice that a few samples have more than one entry. Those are samples that were sequenced a second time to improve coverage. They will be merged during the make.contigs step of mothur.

At this point we need to run code/mothur.batch, which contains most of the commands we'll run in mothur. The mothur.pbs script can be used to run mothur.batch on a high-performance computing cluster. You will likely need to make modifications to mothur.pbs to run it on your cluster.

qsub code/mothur.pbs

OTU clustering and classification

The next step is OTU clustering (and some downstream mothur commands) using the cluster.batch file. This step is very memory intensive, so we run it with 1 processors and 150gb of memory.

qsub code/cluster.pbs

Generating and Testing models

The bulk of the analyses for this study are done through the Baxter_glne007_2016_manuscript.Rmd knitr document. It will generate and test the random forest models, generate figures, and render a docx of the manuscript.

Rscript -e "library(knitr); knit('Baxter_glne007_2016_manuscript.Rmd')"