zty2009/GCN-DNN

cannot generate the file sd_drugfeature46.txt

Opened this issue · 4 comments

this is an empty file

In drugnet_and_feature.R :'catename=as.character(a[which(a[,2]>100),1])'
means that The drug category can be used as a feature if it repeats more than 100 times.
In this demo, the number of drugs is too small, so sd_drugfeature46.txt can not be generated. You can use st_drugfeature27.txt to replace it.
The 25/27 categories used in this paper could be found in supplementary file.

In drugnet_and_feature.R :'catename=as.character(a[which(a[,2]>100),1])'
means that The drug category can be used as a feature if it repeats more than 100 times.
In this demo, the number of drugs is too small, so sd_drugfeature46.txt can not be generated. You can use st_drugfeature27.txt to replace it.
The 25/27 categories used in this paper could be found in supplementary file.

Do you mean using the st_drugfeature27.txt to replace sd_drugfeature46.txt ? If so, In the sd_demo.R, I met the problem:
image

In drugnet_and_feature.R :'catename=as.character(a[which(a[,2]>100),1])'
means that The drug category can be used as a feature if it repeats more than 100 times.
In this demo, the number of drugs is too small, so sd_drugfeature46.txt can not be generated. You can use st_drugfeature27.txt to replace it.
The 25/27 categories used in this paper could be found in supplementary file.

Do you mean using the st_drugfeature27.txt to replace sd_drugfeature46.txt ? If so, In the sd_demo.R, I met the problem:
image

I tried to use the st_drugfeature27.txt to replace sd_drugfeature46.txt then I met the same problem as the one above. In addtion, I try to use the Yamanishi_08 data' s data to replace demodata.txt, and I met the new problem:
image

I just tested my code and found both sp_demo and st_demo can be run. Since the dimension of drugs features in SD test is different from SP and ST, slight change in code is needed. However, the process and method are the same for the three tests. Therefore, you can ignore this problem and use the core part of our method to test on your datasets.
I took a look at your error message, which is a programming problem. Obviously, I can't guide you line by line to write code under different data sets. When you try to use GCN-DTI on another datasets, slight change in code is necessary.
If you come across problems about algorithm(GCN-DTI), please contact me with no hesitate.