AutoReportLite

AutoReportLite is an automatic analysis pipline and reporting system. It will parse the output of your custom analysis pipeline and create a multi-page LaTeX beamer style report, aggregating figures, sample sheets, and other text or image based files into a single document.

Example Report Output

Installation

To install AutoReportLite, simply clone this repository with the following terminal command:

git clone https://github.com/stevekm/AutoReportLite.git

Setup Your Pipeline

Step 1.

Create sample sheet to describe samples for analysis and other relevant information per sample (sample-sheet.tsv)

Step 2.

Setup analysis pipeline script using items from sample sheet (code/analysis_pipeline.sh)

Step 3.

Parse the sample sheet to create subdirectories per sample and pass items to the script for analysis (workflow.Rmd)

Compile the Report

Step 4.

Adjust project_info.txt file to set project information, file paths, figures, as needed for inclusion in the report.

Step 5.

Compile the report.Rnw file with knitr:

$ # in terminal, change to report dir and start R
$ cd report
$ R
> # if knitr is not installed, install it:
> # install.packages("knitr")
> # load knitr
> library("knitr")
> # 'knit' the file
> knit("auto-report.Rnw")
...
output file: auto-report.tex

[1] "auto-report.tex"
> # quit R and return to the terminal
> quit()

Step 6.

Compile the resulting TEX file with pdflatex 2 or 3 times for full compilation.

$ pdflatex auto-report.tex && pdflatex auto-report.tex && pdflatex auto-report.tex

TIPS:

• Use the workflow.Rmd to set up your pipeline's directory structure
• Create a parent directory for the pipeline output (e.g. analysis_output), with a subdirectory for each sample to be processed (Sample1,Sample2,etc.); the name of the subdirectory should correspond to the name or ID for the sample. See this example:

steve@macbook:~/AutoReportLite$ tree analysis_output/
analysis_output/
|-- Sample1
|   |-- R_stats.txt
|   |-- Sample1.distribution_histogram.pdf
|   |-- Sample1.random_distribution.pdf
|   `-- logs
|       |-- analysis_pipeline.sh.e4783069
|       |-- analysis_pipeline.sh.o4783069
|       |-- analysis_pipeline.sh.pe4783069
|       |-- analysis_pipeline.sh.po4783069
|       `-- scriptlog.analysis_pipeline.sh.20160429t173756.highmem001.61335.15519
|-- Sample2
|   |-- R_stats.txt
|   |-- Sample2.distribution_histogram.pdf
|   |-- Sample2.random_distribution.pdf
|   `-- logs
|       |-- analysis_pipeline.sh.e4783070
|       |-- analysis_pipeline.sh.o4783070
|       |-- analysis_pipeline.sh.pe4783070
|       |-- analysis_pipeline.sh.po4783070
|       `-- scriptlog.analysis_pipeline.sh.20160429t173756.highmem001.61390.15152

...

• Set up your pipeline script to work on one sample, and output all of that sample's results in its corresponding pipeline subdir
• If not using qsub for script submission then pipe the stdout and stderr from the pipeline script to log files
• If not using qsub then run the pipeline script in a for loop from the workflow.Rmd
• You can easily embed the code for your analysis pipeline directly into the report RNW file; I omit this here since my pipelines often require resources unavailable in this context, and you may run into problems with qsub job output dependencies and background processes.

---

Dependencies

• R programming language (developed under R version 3.3.0, but older versions will likely work as well)
• knitr R package
• LaTeX and pdflatex (developed under pdfTeX 3.14159265-2.6-1.40.16 (TeX Live 2015), but others may work)
• bash (developed under GNU bash, version 4.1.2(1)-release (x86_64-redhat-linux-gnu), but others may work)
• any software used by your specific pipelines