robertaboukhalil / ginkgo

Cloud-based single-cell copy-number variation analysis tool

Ginkgo

Ginkgo is a cloud-based single-cell copy-number variation analysis tool.

Launch Ginkgo: qb.cshl.edu/ginkgo

Usage

• Step 0: Upload .bed files
• Step 1: Choose analysis parameters
• Step 2: Compute Copy Number Profiles, and a Phylogenetic Tree
• Step 3: Analyze Individual Cells

Setup Ginkgo on your own server

Requirements:

• PHP >=5.2
• R >= 3.0.0
• R Packages:
  • ctc
  • DNAcopy
  • inline
  • gplots
  • scales
  • plyr
  • ggplot2
  • gridExtra
  • fastcluster
  • heatmap3

WARNING Version 3.0.0 (Mar 28, 2016) of gplots introduced a bug in heatmap.2 that makes it calculate dendrograms even when Rowv or Colv is set to FALSE so that Ginkgo will run for a very long time. The solution is to use an older version of gplots or use fixed version from: https://github.com/ChristophH/gplots. This can be installed using the commands below

remove.packages('gplots'); 
library('devtools'); 
install_github("ChristophH/gplots")

Install Ginkgo:

Type make in the ginkgo/ directory

Server Configuration:

• /etc/php.ini

  • upload_tmp_dir: make sure this directory has write permission
  • upload_max_filesize: set to >2G since .bam files can be large

• ginkgo/includes/fileupload/server/php/UploadHandler.php

  • In constructor, on line 43 and 44:
    • upload_dir = [FULL_PATH_TO_UPLOADS_DIR] . $_SESSION["user_id"] . '/'
    • upload_url = [FULL_URL_TO_UPLOADS_DIR] . $_SESSION["user_id"] . '/'

• ginkgo/bootstrap.php

  • Change DIR_ROOT, DIR_UPLOADS and URL_ROOT

• ginkgo/scripts/analyze.sh

  • Change home variable to where the ginkgo/ folder is located

• ginkgo/scripts/process.R

  • Change main_dir variable to the folder where ginkgo/scripts is located

• ginkgo/scripts/reclust.R

  • Change main_dir variable to the folder where ginkgo/scripts is located

• ginkgo/scripts/analyze-subset.R

  • Set the folder to where ginkgo/scripts is located
  • Set the folder to where ginkgo/genomes is located (warning: test this carefully if your ginkgo/uploads folder is a symlink)

• Make sure the uploads directory has the correct write permissions

Download data files:

• Download binning data for hg19 at https://labshare.cshl.edu/shares/schatzlab/www-data/ginkgo/genomes/hg19.tgz

  • untar into ginkgo/genomes/hg19 (which needs to be created)

• Other genomes, including hg18, hg19, mm9, mm10, rheMac7, rheMac8, rn5 and dm3 can be found at http://labshare.cshl.edu/shares/schatzlab/www-data/ginkgo/genomes/