DeepTIL

DeepTIL (Deep deconvolution of Tumor Infiltrating Leucocytes) quantifies the leukocyte composition of a sample. To achieve this, it combines Cibersort leucocyte fractions and total leukocyte abundance, calculated as normalized enrichment scores by AutoCompare_SES_norm.

DeepTIL can use either Cibersort-LM22 or Cibersort-LM7 to quantify the respectively 22 or 7 leukocyte subsets. The leukocyte abundance in a sample is computed by its sample enrichment score (SES) as depicted in ref. Briefly SES returns the -log(Pvalue) for testing (gene set rank distribution is greater than uniform distribution). To account for gene sets of different sizes, the raw SES are normalized on a 0-100 scale with 100 as maximal theoretical SES for the gene set size.

Citation :

When using DeepTIL to process data for publication please cite:

Assessment of tumor-infiltrating TCRVγ9Vδ2 γδ lymphocyte abundance by deconvolution of human cancers microarrays.

Marie Tosolini, Frédéric Pont, Mary Poupot, François Vergez, Marie-Laure Nicolau-Travers, David Vermijlen, Jean- Emmanuel Sarry, Francesco Dieli, Jean-Jacques Fournié . OncoImmunology , 2017 : 6;6(3).

Manual :

Single thread :

1- install Julia 1.7 or later (command line version) free programming language and install package :

CSV : add CSV

DataFrames : add DataFrames

ArgParse : add ArgParse

2- unzip the software

3- copy your Cibersort tables files in the “cibersort” directory.

- txt or csv files with TAB separator by default

- first line : column names

- first column : sample names (sample names must be strings, not numbers, othewise it produces a type error, Int instead of string)

- next columns : values with dot as decimal separator.

exemple :

4- copy your SES tables files in the “ses" directory. Tables must have ".tsv" extension. SES have to be calculated using the LM7 file as database in AutoCompare_SES_norm. Just copy the LM7 file in <Autocompare_SES_norm/databases/MyGeneSets/> for example, run Autocompare_SES_norm and get the SES files in <Autocompare_SES_norm/results_norm/>.

- txt or csv files with TAB separator by default

- first line : column names must contain the 7 folowing names : B, CD4, CD8, granulocytes, MoMaDC, NK, Tgd. Only those columns will be used for calculation.

- first column : sample names

- next columns : values with dot as decimal separator.

example :

5- execute the software by the command :

julia deeptil-005.jl -f CIBERSORT_file -e SES_file

example :

julia deeptil-005.jl -f cibersort/CIBERSORT_LM7_GSEXX.txt -e ses/SES_LM7_GSEXX.tsv

6- the results are in the results directory.

Parallel file processing (linux only) :

This version is usefull to process many Cibersort and SES files in parallel.

Same installation and procedure than for single CPU.

1- install GNU parallel and Perl

2- very important Cibersort and SES file names must be formated as follows :

CIBERSORT.samplename.txt

SES_samplename_norm.tsv

do not remove the dot after CIBERSORT !

3- execute the software by the command :

perl parallel_deeptil-05.pl nb_of_cpus

example :

perl parallel_deeptil-05.pl 10 # parallel computing of 10 files on 10 cpus

4- the results are in the results directory.