DeepAllodriver Help

Introduction

The recent advances in next-generation sequencing technologies and relative databases have allowed the cancer researchers to use more computational methods to study driver mutations and gene-expressed mutated proteins in which some of them have become successful drug targets. Due to the advantages of the protein target at allosteric sites, the recognition of allosteric protein targets has become a new direction of cancer therapy.

Our former pan-cancer analysis for 7000 tumor-normal matched samples revealed that somatic deleterious mutations are more significantly enriched at allosteric and orthosteric site than other regions on protein. Inspired by it, we collected the clinical mutational samples from TCGA, ICGC as well as COSMIC database and mapped mutations to orthosteric sites/ allosteric sites and potential allosteric site derived from protein structures. Then we constructed a subgraph based on the mapped mutational residues and its neighbors residues, ground on our previously research and method, we developed a web server called DeepAlloDriver which used a graph convolutional network model to predict the driver ability of mutations. The DeepAlloDriver provides a functional protein sites perspective to help researchers to find novel drug targets based on the somatic mutations.

Software requirement

The DeepAlloDriver requires a modern web browser with JavaScript enabled. To view the submission status of a job, pop-ups must not be blocked. The following browsers have been thoroughly tested with DeepAlloDriver:

Mozilla Firefox, version 5 or above(Download);

Chrome, version 9 or above(Download);

Safari, version 5 or above(Download);

Edge, version X or above(Download);

The latest version of Firefox and Chrome is recommended for visualization.

Below are the detailed operation system and brower version we have tested that our website can opertate successfully.

OS	Version	Chrome	Firefox	Microsoft Edge	Safari
Windows	10	109.0.5414.120	109.0	109.0.1518.78	not tested
Windows	10	99.0.4844.51	109.0	109.0.1518.78	not tested
Windows	10	98.0.4758.82	109.0	109.0.1518.70	not tested
Windows	7	108.0.5359.125	103.0.1	109.0.1518.78	not tested
MacOS	Monterey	109.0.5414.119	109.0.1	109.0.1518.78	15.1 (17612.2.9.1.20)
Linux	CentOS	109.0.5414.119	n/a	107.0.1418.52	not tested
Linux	Ubuntu	n/a	n/a	n/a	n/a

Tutorials

We would introduce how to use and how to analyze DeepAlloDriver with the example below.

Tutorials Table
Method	Input Format & Examples	Run Time	Tutorial
Example1	TSV: sample1;AKT1;E17K sample2;ESR1;E385Q sample3;FLT3;F691L sample3;ABL1;Y449C sample4;ACVR1;L495F sample5;AR;W742L	Less than 2 minute	View
Example2	TSV: sample1;IDH1;M291I sample2;IDH1;A258T sample2;SMO;W281C sample2;ABL1;Y449C sample2;ACVR1;R375C sample2;ACVR1;S192N	Less than 2 minute	View

Introduction of Example1

Our mutations of example derive from COSMIC (COSMIC, the Catalogue Of Somatic Mutations In Cancer, is the world's largest and most comprehensive resource for exploring the impact of somatic mutations in human cancer). Here is the general overview of the selected mutation.

Sample ID	Source WebSite	Source Identifer	Source Link
sample1
sample2

Step 1: DeepAllodriver Input of Example1

Click Home of DeepAllodriver menu, and press the Example1 button. DeepAllodriver will automatically load parameters for the user in the Job Submit form:

1. Input the Job Name "DeepAllodriver_EXAMPLE1";

2. Paste your mutation content into the Input box;
Step 2: Job Submitting

In the case of examples, all parameters and selection are set automatically and can not be changed. When clicking Run button, the job automatically begins to run.
Step 3: DeepAllodriver Result

It takes ~2min for Example to identify the targets of clinical mutations. The Job Process section has recorded the whole progress of the job running.

When Example is finished, all results are shown in the panels of Job Queue page. The results can be downloaded by clicking the two buttons with different kinds of files.

The panels provides the detailed information of the driver mutations and their targets.

In the Target Result panel, predicted targets are listed with eight features including Sample ID,Gene/Protein, UniProt ID, Mutation, PDB ID, Predicted Score and Entry.

DeepAlloDriver successfully predicts the potential target based on input mutations.

Clicking Show button on each mutation record may open a new page with more details, which will be introduced in Step 4.
Step 4: Browse details of driver mutation

Clicking Show button on each mutation record in Step 3 may open a new page with more details. This page contained four important parts:

1. Target information:

In the Target information panel, a window in the left part displays the 3D structure of the protein by using JSMol plugin,Visualization of mutations on 3D structures. The mutated residue is displayed as spheres.

And in the right part, there is a table of general information of the protein and its mutation information, such as Gene Symbol, NCBI Gene ID, Function, Ensembl ID, PDB ID,and Predicted Score.

Hover on the Function to get details of gene functions.

2. Pathway in Reactome:

The table indicates the pathway of Reactome database which the target participate in.

3. Known drugs on the target:

The table indicates the known therapeutic modulators targeting this protein in DrugBank and CHEMBL database.

- Introduction of Example2
  
  Our mutations of example derive from COSMIC (COSMIC, the Catalogue Of Somatic Mutations In Cancer, is the world's largest and most comprehensive resource for exploring the impact of somatic mutations in human cancer). Here is the general overview of the selected mutation.
- Step 1: DeepAllodriver Input of Example2
  
  Click Home of DeepAllodriver menu, and press the Example2 button. DeepAllodriver will automatically load parameters for the user in the Job Submit form:
  
  1. Input the Job Name "DeepAllodriver_EXAMPLE2";
  
  2. Paste your mutation content into the Input box;
- Step 2: Job Submitting
  
  In the case of examples, all parameters and selection are set automatically and can not be changed. When clicking Run button, the job automatically begins to run.
- Step 3: DeepAllodriver Result
  
  It takes ~2min for Example2 to identify the targets of clinical mutations. The Job Process section has recorded the whole progress of the job running.
  
  When Example is finished, all results are shown in the panels of Job Queue page. The results can be downloaded by clicking the two buttons with different kinds of files.
  
  The panels provides the detailed information of the driver mutations and their targets.
  
  In the Target Result panel, predicted targets are listed with eight features including Sample ID,Gene/Protein, UniProt ID, PDB ID, Predicted Score and Entry.
  
  DeepAlloDriver successfully predicts the potential target based on input mutations.
  
  Clicking Show button on each mutation record may open a new page with more details, which will be introduced in Step 4.
- Step 4: Browse details of driver mutation
  
  Clicking Show button on each mutation record in Step 3 may open a new page with more details. This page contained four important parts:
  
  1. Target information:
  
  In the Target information panel, a window in the left part displays the 3D structure of the protein by using JSMol plugin,Visualization of mutations on 3D structures. The mutated residue is displayed as spheres.
  
  And in the right part, there is a table of general information of the protein and its mutation information, such as Gene Symbol, NCBI Gene ID, Function, Ensembl ID, PDB ID, and Driver Gene.
  
  Hover on the Function to get details of gene functions.
  
  2. Pathway in Reactome:
  
  The table indicates the pathway ofReactome database which the target participate in.
  
  3. Known drugs on the target:
  
  The table indicates the known therapeutic modulators targeting this protein in DrugBank and CHEMBL database.

Input Form

Click Home in the DeepAllodriver menu. Fill in the parameters in the Job Submit form:

The descriptions of the input parameters are listed in the following table:

Input Parameters

No.	Parameter	Detail
1	Job Name	User must input an unique name for new Job.
2	Input	User may input or upload sequencing profile of cancer samples annotated with mutation data. Three file formats are allowed: TXT, MAF, and ANNOVAR.

Input File Format

1.TXT:paste your mutation content into the text area:

The input mutation list contains three columns: Sample ID, Gene name and Mutation, which are separated by semicolon. The format are as following image:

ATTENTION: The input contents for prediction should be no more than 10M.

2. MAF: click the MAF button and browse your MAF file. Then, click the Browse button for file selecting. The file extension is allowed to be maf or txt. The file will be uploaded automatically.

A Mutation Annotation Format (MAF) file (.maf) is a human-readable and tab-delimited text file that lists mutations derived from sequencing cancer cohort samples. The format originates from The Cancer Genome Atlas (TCGA) project and has been widely accepted by researchers since the detected mutations in a MAF file can be easily viewed, edited or processed for custom mining. Each line in a MAF file represents a cancer mutation, which is annotated by many fields ranging from chromosome names to protein information. Column headers and ordering may sometimes vary between files of different sources.

In DeepAlloDriver, the MAF file is composed of four headers as follows: (1) "Hugo_Symbol" for gene symbol, e.g. BRAF; (2) "HGVSp_Short" for detected mutation, e.g. p.V600E; (3) "Variant_Classification" for translational effect of variant allele, e.g. Missense_Mutation; (4) "Tumor_Sample_Barcode" for cancer sample ID. More headers in MAF file are described in the GDC documentation of NIH (https://docs.gdc.cancer.gov/Data/File_Formats/MAF_Format/ ).

3. ANNOVAR: click the ANNOVAR button and browse your ANNOVAR file. Then, click the Browse button to choose a file. The file extension is allowed to be annovar. The example annovar file can be downloaded here. The file will be uploaded automatically.

ANNOVAR is a widely-used software to annotate genetic variants detected from diverse genomes. In DeepAlloDriver, a tab-separated annotation output file from ANNOVAR software are supported. The variant annotation file can be generated by the table_annovar.pl script in the program.

ATTENTION:The input files for prediction should should be no more than 10M.

Submit the Job

After pasting or uploading the file, the user could submit the parameters by click the Run button.
Reset the Form
Click Reset button to reset the form.

Warning of Input

1. Job Name empty warning: the blank Job Name parameter will get a warning pop-up message. The user should specify the Job Name in the form.

2. TXT empty content warning: the blank or TXT text area will get a warning pop-up message as following picture. The user should input the mutation content or choose other file format.

- Result Download
  
  When the job is finished, all results are shown in the panels of Job Queue page. The results can be downloaded by clicking the two buttons with different kinds of files.
- Target Result
  
  In the Target Result panel, top 100 ranked driver mutations are listed with eight features including Sample ID,Gene/Protein, UniProt ID, Mutation, PDB ID, Predicted Score and Entry.
  
  DeepAlloDriver successfully predicts the potential target based on input mutations.
  
  Clicking Show button on each mutation record may open a new page with more details, which will be introduced in Step 4.
- Browse details of driver mutation
  
  Clicking Show button on each mutation record in Step 3 may open a new page with more details. This page contained four important parts:
  
  1. Target information:
  
  In the Target information panel, a window in the left part displays the 3D structure of the protein by using JSMol plugin,Visualization of mutations on 3D structures. The mutated residue is displayed as spheres.
  
  And in the right part, there is a table of general information of the protein PTPN11 and its mutation information, such as Gene Symbol, NCBI Gene ID, Function, Ensembl ID, PDB ID, Mutation, and Predicted Score.
  
  Hover on the Function to get details of gene functions.
  
  2. Pathway in Reactome:
  
  The table indicates the pathway of Reactome database which the target participate in.
  
  3. Known drugs on the target:
  
  The table indicates the known therapeutic modulators targeting this protein in DrugBank and CHEMBL database.

Job Search

Click Job Queue to show Job Search page and here user can retrieve his/her own jobs. By inputting the type of "Job ID" and unique identifier from user in the "KeyWord", jobs of the user could be listed in a pop-up tab.

ATTENTION: Message "No hits found" will emerge if identifier from user can not be exactly matched.

Job Status

Job Progress panel shows the job progress after submitting a job.

The status of the jobs is refreshed in the panel every 10 seconds till "Done" emerges. When job is finished, the Job progress panel shrinks back and the result of the job is automatically shown in the Job Result panel. The user can bookmark this page for further analysis by clicking the Bookmark the Result.

An DeepAllodriver job will take about 1-60 minutes, which depends on the number of clinical samples.

ATTENTION: if Error number appears, a warning message pops up to help evaluating the error type according to the following codes.

Job Running Error

No.	Status	Detail & Solution
1	Error. No mutations mapped to the deposited proteins.	No clinical mutations in uploaded samples mapped to the annotated proteins deposited in DeepAllodriver.
2	Error. No mutations mapped to the allosteric sites.	No clinical mutations in uploaded samples mapped to the allosteric sites in DeepAllodriver.
3	Calculation Error.	Calculation Failed. Please contact us.

DeepAllodriver

Introduction

Software requirement

Tutorials

Contact us