I. DIGREM input instructions

1. Drug-treated gene expression data
Gene expression profiles of single compound-treated and negative control-treated cell line sample. Data needs to be log transformed.

2. Dose-response curve data (optional)
Dose-response curves for viability of the cell lines you used for gene expression assay. For each compound, two data points(viability reduction in IC20 and double dose of IC20) from the dose-response curve are desired.

3. Pathway information
Pathway information is used to estimate the drug similarity of effect on upstream and downstream genes. The KEGG pathway is a universal gene network database, but it is not context-specific. Users can choose to upload their own gene-gene interaction information if their data is generated on specific cancer cell lines that have a unique gene regulatory relationship.

By taking the above three inputs, DIGREM will calculate the pair synergistic scores of all possible combinations of the compound provided, and their rank from the most synergistic to the most antagonist. A larger score indicates a higher possibility for the pair to have synergistic effect, and vice versa.

II. DIGREM methods instruction

The DIGREM will automatically call three methods to do the prediction if the requirements are met, and will output the predicted result of each method.


1. DIGRE model
DIGRE is the method with the fewest requirements, and therefore it is always run. It takes the above-mentioned three inputs to do the prediction with tolerance of no dose response data. See Details

2. IUPUI_CCBB model
The IUPUI_CCBB method requires the gene expression data to have at least three replicates for each drug (to ensure enough statistical power). If the requirement is met, this method will be run, and the prediction will be added to the final output report. Otherwise this method is skipped. See Details

3. Gene set-based model
The gene set-based method requires the gene expression data to contain at least 16 columns to do normalization. If the requirement is met, this method will be run, and the prediction will be added to the final output report. Otherwise this method is skipped. See Details

4. Correlation-based model
The correlation-based method also requires relatively large gene expression data matrix (at least 16 drugs or replicates). If the requirement is met, this method will be run. Based on our experience, this method has the least consistency with the other three methods, we suggest fousing more on prediction of the other three methods. See Details

III. DIGREM outputs instruction

1. Heatmap
A heatmap of DIGRE-predicted synergistic scores. Dark blue represents high predicted synergy value, and white represents low synergy value. Only the upper triangle of the results is displayed.

2. Bar plot
Bar plot of compound pairs with top DIGRE synergistic scores.

3. Download file
Raw predicted score of all compound pairs by DIGRE, IUPUI_CCBB and gene set-based method.

Please leave your contact information (optional).
We will send you email after work done.




* Methods Executed:

 DIGRE model
 IUPUI_CCBB model
Gene set-based model
 Correlation-based model

* Gene Expression data:

Only accept CSV fil. Maximum file size 10M.
The first column must be gene names. The first row of other columns should be compound name.
At least one column of negative control data should be provided. And the column name should be "Neg_control".
Same compound may have more than one column, in which case the average values will be use in calculation.

Example:

  Dose data (optional):

Only accept CSV fil. Maximum file size 60K.
The content should have three rows.
The first row should be compound names. Each compound should only have one column.
The second row should be the values of doseRes1. The first column should be labeled as "doseRes1".
The third row should be the values of doseRes2. The first column should be labeled as "doseRes2".

Example:

* 

* Connectivity data:

Only accept CSV file. Maximum file size 1M.
The file should have two columns to specify the two connected genes with their SYMBOL name.
Example: the following gene-network example files are refined from lymphoma patients gene expression data. If you are using lymphoma cell lines, you could use our gene network, otherwise we recommend you use KEGG pathway or upload your own gene-network file.

Gene network: