SNUG

SNUG is a single-sample network-based drug efficacy prediction framework that improves traditional network proximity analysis by integrating individual signatures (IS), high-confidence single-sample networks (hcSINs) and proximity-based hypergeometric test (PHT) scoring.

The SNUG workflow consists of four main steps:

• Step 1: single-sample networks (SINs) construction
• Step 2: high-confidence single-sample networks (hcSINs) construction
• Step 3: individual signature (IS) selection
• Step 4: proximity-based hypergeometric test (PHT)

Dependencies

The code was developed and tested with the following software:

• Python 3.14.8
• NumPy 2.4.3
• SciPy 1.17.1
• NetworkX 3.6.1
• kneed 0.8.5

Install the required packages using:

pip install "package name"

Input File Format

Gene expression matrix (GEM) file

The input GEM file should be a tab-separated .txt file with the following structure:

gene id	sample 1	sample 2	...	sample n
gene 1	...	...	...	...
gene 2	...	...	...	...
...	...	...	...	...
gene n	...	...	...	...
...	...	...	...	...

Note: Gene IDs should be converted to Entrez Gene IDs before running SNUG.

Background network file

The input network file should be a tab-separated .txt file with the following columns:

Entrez1 Entrez2 protein1 protein2 combined_score

Drug target file

The input drug target file should be a .txt file in the following format (without a header). Each row should contain a drug name followed by its target genes.

drug A    target 1;target 2;...
drug B    target 1;target 2;...
drug C    target 1;target 2;...
...

Note:

• Target genes should be converted to Entrez Gene IDs before running SNUG.
• The targets for each drug should be separated by ;.

Basic Usage

The demo dataset is a small example for testing the workflow and does not represent the complete dataset used in the manuscript.

Step 1: single-sample networks (SINs) construction

Step 1-1: calculate genome-wide sample weights:

python3 ./SWEET/1-1_correlation_to_weight.py -g ./demo/demo_GEM.txt -c ./demo/demo_correlation.txt -w ./demo/demo_weight.txt

-h: Get help with the commands.
-g: Gene expression matrix (GEM) file.
-c: Output file containing sample to sample correlation matrix.
-w: Output file containing weight of each sample.

Step 1-2: Calculate the mean value of all samples.

python3 ./SWEET/1-2_sweet_mean.py -g ./demo/demo_GEM.txt -w ./demo/demo_weight.txt -o ./demo/demo_mean.txt

-h: Get help with the commands.
-g: Gene expression matrix (GEM) file.
-w: Sample weight file (i.e., the output file from step 1-1).
-o: Output file containing mean value of all samples.
--k: Balance parameter (default: 0.1).

Step 1-3: Calculate the standard deviation (std) of all samples.

python3 ./SWEET/1-3_sweet_std.py -g ./demo/demo_GEM.txt -w ./demo/demo_weight.txt -m ./demo/demo_mean.txt -o ./demo/demo_std.txt

-h: Get help with the commands.
-g: Gene expression matrix (GEM) file.
-w: Sample weight file (i.e., the output file from step 1-1).
-m: Sample mean value file (i.e., the output file from step 1-2).
-o: Output file containing standard deviation of all samples.
--k: Balance parameter (default: 0.1).

Step 1-4: Construct SINs of each sample.

python3 ./SWEET/1-4_sweet_degree_split_output_network.py -g ./demo/demo_GEM.txt -w ./demo/demo_weight.txt -m ./demo/demo_mean.txt -s ./demo/demo_std.txt -o ./demo/demo_SIN

-h: Get help with the commands.
-g: Gene expression matrix (GEM) file.
-w: Sample weight file (i.e., the output file from step 1-1).
-m: Sample mean value file (i.e., the output file from step 1-2).
-s: Sample standard deviation file (i.e., the output file from step 1-3).
-o: Path to the folder containing SIN files.

Step 2: high-confidence single-sample networks (hcSINs) construction

python3 ./hcSIN/2_hcSIN_construction.py -n ./demo/demo_network.txt -w ./demo/demo_weight.txt -s ./demo/demo_SIN -hc ./demo/demo_hcSIN

-h: Get help with the commands.
-n: Background network file.
-w: Sample weight file (i.e., the output file from step 1-1).
-s: Path to the folder containing SIN files (i.e., the output file from step 1-4).
-hc: Path to the folder containing hcSIN files.

Step 3: individual signature (IS) selection

python3 ./IS/3_individual_signature_selection.py -w ./demo/demo_weight.txt -hc ./demo/demo_hcSIN -o ./demo/demo_individual_signature.txt

-h: Get help with the commands.
-w: Sample weight file (i.e., the output file from step 1-1).
-hc: Path to the folder containing hcSIN files (i.e., the output file from step 2).
-o: Output file containing individual signatures of each sample.

Step 4: proximity-based hypergeometric test (PHT)

python3 ./PHT/4_PHT.py -hc ./demo/demo_hcSIN -i ./demo/demo_individual_signature.txt -d ./demo/demo_drug_target.txt -o ./demo/demo_SNUG.txt

-h: Get help with the commands.
-hc: Path to the folder containing hcSIN files (i.e., the output file from step 2).
-i: Individual signature file (i.e., the output file from step 3).
-d: Drug target file.
-o: Output file containing the PHT results for each sample.