LINCS Extraction

Extraction of LINCS L1000 data.

A Next Generation Connectivity Map: L1000 platform and the first 1,000,000 profiles

Repo image from The Library of Integrated Network-Based Cellular Signatures NIH Program: System-Level Cataloging of Human Cells Response to Perturbations

Authors: Maria-Anna Trapotsi ([email protected]) and Layla Hosseini-Gerami ([email protected])

Before you start, make sure you have the following packages installed. We recommend you create a Conda environment for this purpose.

• pandas
• numpy
• cmapPy
• h5py
• scipy
• joblib

Step 1: Create folders

Create folders called "data", "Consensus_Signatures", "TAS_Scores", "Compounds_With_No_Replicates", "All_Replicates" in your working directory

Working Directory/
├── data/
│   ├── GSE92742_Broad_LINCS_Level5_COMPZ.MODZ_n473647x12328.gctx.gz
│   ├── GSE92742_Broad_LINCS_gene_info.txt.gz
│   ├── GSE92742_Broad_LINCS_sig_info.txt.gz 
│   ├── GSE70138_Broad_LINCS_Level5_COMPZ_n118050x12328_2017-03-06.gctx.gz
│   └── GSE70138_Broad_LINCS_sig_info_2017-03-06.txt.gz
├── Consensus_Signatures/
├── TAS_Scores/
├── Compounds_With_No_Replicates/
├── All_Replicates/
├── automation.sh
└── Extract_And_TAS.py

Step 2: Extract LINCS L1000 files from Phase 1 and Phase 2 from Gene Expression Omnibus (GEO) and save them in "data" folder

For Phase 1 visit: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE92742 and download the following files: GSE92742_Broad_LINCS_Level5_COMPZ.MODZ_n473647x12328.gctx.gz GSE92742_Broad_LINCS_sig_info.txt.gz GSE92742_Broad_LINCS_gene_info.txt.gz

For Phase 2 visit: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE70138 and download the following files: GSE70138_Broad_LINCS_Level5_COMPZ_n118050x12328_2017-03-06.gctx.gz GSE70138_Broad_LINCS_sig_info_2017-03-06.txt.gz

Step 3: Run the Extract_And_TAS.py

Extract_And_TAS.py options:

Options:
  -h, --help            show this help message and exit
  --cell_line=CELL_LINE
                        Cell line for extraction (default MCF7)
  --pert_time=PERTURBATION_TIME
                        Perturbation time for extraction (hours) (default 24)
  --pert_dose=PERTURBATION_DOSE
                        Perturbation dose for extraction (uM) (default 10)
  --ncores=CORES        Number of cores (default 1)

I.e. if you want to extract all data from VCAP measured in 24 hours and 5 uM, using 5 cores of compute power you would type:

$ python Extract_And_Tas.py --cell_line=VCAP --pert_time=24 pert_dose=5 --ncores=5

Please note you may have to change the filepaths in the code to match with the filenames you have extracted in data/ folder, as filenames are periodically updated.

This script will extract LANDMARK GENES ONLY but you can modify a couple of lines in the code to extract best inferred (BING) or all genes also. Please get in touch if you need help with this.

automation.sh will enable you to extract data in multiple conditions, by providing a few parameters which will be looped over in every combination - you may edit this file to keep only cell lines etc. you care about.

$ chmod +x automation.sh # make it executable
$ ./automation.sh

Extracted Data

1. Compounds with replicates: Consensus signatures are saved in the full matrix of all compounds in conditions specified (e.g A375_24 h_10uM.txt) where headers are genes and rows are compounds, in the "Consensus_Signatures" folder.

Consensus signatures are computed for biological replicates in the same way that the original paper details for technical replicates.

Individual replicates are also extracted; these are saved as a matrix per compound (e.g. compoundname_A375_24 h_10uM.txt) where rows are genes and headers are compound replicate ID, in the "All_Replicates" folder.

2. Compounds with no replicates: Individual signatures are saved in the full matrix of all compounds in conditions specified (as above) but in the "Compounds_With_No_Replicates" folder.

3. TAS scores for all compounds with replicates TAS score + SS and CC scores for all compounds with replicates in conditions specified (e.g. A375_24 h_10uM_TAS.txt) in TAS_Scores folder.

TAS scores are computed for biological replicates in the same way that the original paper details for technical replicates.

!!!If you just want a matrix of gene expression data then you will need to combine the matrix in Consensus_Signatures and Compounds_With_No_Replicates for your condition of interest (simple Python concat function)!!!

Future functionality

1. Ability to provide more than one cell line, time point, dose etc. as an option (for now automation.sh script allows iteration, but each extraction is a separate task i.e. not as a loop within the Python script)
2. Command line option to choose gene subset (LM, BING, All - for now just LM but this can manually be edited in the Python script)
3. Command line option for TAS score calculation or not