Requirement

[yingh...@hotmail.com]

Hi author,

I am trying to run Hotnet2 to find subnetworks a disease, I got two files now. The first one is the heat score containing the associate score  from Open Target for Alzheimers, and the second one is the PPI network from STRING database as shown. Before running HotNet2.py, I need to run makeNetworkFiles.py, but the output file for that has no gene names in it, then how I can make sure the association score can be attached to the right gene while running HotNet2.py? Please suggest. 

Thanks,

Helen

 

[matthe...@brown.edu]

Hi Helen,

HotNet2 does not convert between different gene or protein identifers, e.g., HUGO symbols and Ensembl protein IDs.  In your case, you can convert the Ensembl protein IDs to HUGO gene symbols in your network file so that they match the genes in your heat score file.  To do so, you can use genenames.org or another resource.

Also, in your screenshot, it looks like all of your genes may have the same heat score.  If they do have the same heat score, then HotNet2 may not function as you might expect.  Since HotNet2 identifies significantly altered (hot) subnetworks, it implicitly requires that some genes higher heat scores than others.

Best,

Matt

[yingh...@hotmail.com]

Hi Matthew, 

Thanks for your reply! The heat scores are different for genes in my dataset, they are just in descending order.  

I got some further question regarding the input file, do I have to give all the gene in the PPI a heat score to start? In my current heat file, it contains only scores for some genes in the PPI, do I need to set the score to 0 for those gene has no initial score?

Thanks,

Helen

[matthe...@brown.edu]

Hi Helen,

No, you do not have to give all genes in the PPI network a heat score.  If some genes were not measured, then it is probably better not to assign heat scores to them instead of assigning them heat scores of zero.

Best,

Matt

[yingh...@hotmail.com]

Thanks for your help Matt, I finally got the example code to run.

I got some result files, hint+hi2012_ppr_0.4.h5 could you please give me some idea how to interpret  this file? A screenshot is shown below, what does each column tells me? 

I got some results files for the Hotnet2.py as well,  stats.tsv  subnetworks.json  subnetworks.tsv  viz-data.json, please provide me some general idea how to use them.

Best,

Ying 

[matthe...@brown.edu]

Hi Ying,

The subnetworks.tsv file provides the HotNet2 subnetworks.  Each line contains a different subnetwork.  The subnetworks.json file contains this information as well.  The stats.tsv file describes the statistical significance of HotNet2 subnetworks using the procedure described in the HotNet2 paper.  See Section 1.3 of the supplement for more details.  The viz-data.json file contains data for visualizing the HotNet2 results.

Best,

Matt

[yingh...@hotmail.com]

Hi Matt,

Thanks for your reply.

Just want to to know what are those .json file for? I have never seen this type of file before, and only know that they can be read using python. Also, what software can I use to visualise the results?

Thanks,

Helen 

[matthe...@brown.edu]

Hi Helen,

JSON is a common file format for exchanging data.  The subnetworks.json file uses the JSON file format to save information about the HotNet2 subnetworks.  The subnetworks.tsv file contains much of the information in the subnetworks.tsv file.

Please see the HotNet2 README on GItHub for information about visualizing the HotNet2 subnetworks.  Alternatively, you can use your favorite graph-drawing software, such as Cytoscape, to visualize the HotNet2 subnetworks in the subnetworks.tsv file.

Best,

Matt

[yingh...@hotmail.com]

Hi Matt,

Thanks for your reply.

I got some further question for the makeNetworkFiles. In total, I want to do 50 permutation. If I understand correctly, as there are random network, I can generate them separately can combine the results in permute folder right? As the command line shown below:

python makeNetworkFiles.py \

    -e  Mydata/Omnipath/OmniPath_edgelist.tsv  \

    -i  Mydata/Omnipath/OmniPath_gene_index.tsv \

    -nn Omnipath\

    -p  Omnipath \

    -b  0.42 \

    -o  Mydata/Omnipath \

    -np 20\

    -c  20

python makeNetworkFiles.py \

    -e  Mydata/Omnipath/OmniPath_edgelist.tsv  \

    -i  Mydata/Omnipath/OmniPath_gene_index.tsv \

    -nn Omnipath2\

    -p  Omnipath2 \

    -b  0.42 \

    -o  Mydata/Omnipath \

    -np 25\

    -c  20

But I got a question doing in this way when run HotNet2.py, how should I write the line -pnp so that all random networks can be included? Since they are in different names now, i.e., Omnipath_ppr_0.42_##NUM##.h5, Omnipath2_ppr_0.42_##NUM##.h5.

Best,

Ying 

[matthe...@brown.edu]

Hi Ying,

If you would like to generate permuted networks on different machines, then you can try the following scripts, where the -ps (--permutation_start_index) argument determines the index of the permutation.  In particular, please do not use different network names for the same network on different machines because HotNet2 will think you have different networks.

python makeNetworkFiles.py \

    -e  Mydata/Omnipath/OmniPath_edgelist.tsv  \

    -i  Mydata/Omnipath/OmniPath_gene_index.tsv \

    -nn Omnipath \

    -p  Omnipath \

    -b  0.42 \

    -o  Mydata/Omnipath \

    -ps 1 \

    -np 25 \

    -c  25

    

python makeNetworkFiles.py \

    -e  Mydata/Omnipath/OmniPath_edgelist.tsv  \

    -i  Mydata/Omnipath/OmniPath_gene_index.tsv \

    -nn Omnipath \

    -p  Omnipath \

    -b  0.42 \

    -o  Mydata/Omnipath \

    -ps 26 \

    -np 25 \

    -c  25

Best,

Matt

[yingh...@hotmail.com]

Hi Matt,

Thanks for you reply!

After I got a list of subnetworks, my goal is to calculate the heat score of each gene in a subnetwork. It is there any scrip to do this? If not, would you pleas tell me how to output the E(i,j) matrix? So that I can used it to program myself, or do you have any suggestion to do that?

Thanks,

Ying