PopPhy-CNN source code examples
Example Cirrhosis abundance file download download
T-BAS examples
Example switchgrass fungi T-BAS assignments file download download download
The abundance file is a sample counts table with qiime formatted names. If this is not available then drop the T‑BAS file in the drop box and a sample counts file in the abundance.tsv box.
If the labels file is available as in the Cirrhosis example, then use that. If not then use a samples metadata file with a column named popphy_label that assigns classes to samples
Note that underscores are not allowed in the sample names.
If the names in the first column of the counts table are ASVs, then check this box. If the names in the first column are sample ids then do not check this box.
Filter by Kingdom
This option removes submitted ASVs by kingdom assignmnent. It makes the program run faster if the submission includes multiple kingdoms.
No filter
Filter
T-BAS assignments file.
reset
To use a T-BAS assignments file to create inputs the counts table row and column names must match the assignments file row names and labelled sample attributes file row names, depending on the format of the counts table.
Drop assignments file (optional)
Sample counts table or abundance.tsv
reset
Drop counts table
Sample metadata or labels.txt
reset
Drop labels file
Options help
Options help (L2Lambda)
L2Lambda
If your data is already normalized, a good starting point for the L2 regularization rate (also called "lambda" or "weight decay") is usually a small value like 0.001 or 0.01. Since normalization helps prevent features from dominating due to scale differences, you can apply a slightly stronger regularization penalty compared to unnormalized data.
0.01
0.001
0.0001
KernelWidth
10
5
4
3
KernelHeight
10
5
4
3
NumConvLayers
1
2
3
4
Annotation :
To facilitate combining treescores files from multiple runs add a meaningful annotation.