As I had trouble getting the RawMap compiled with hdf5, I modified it to work with S/BLOW5. Had to comment out all FAST5-related stuff as I wanted to get it working fast, but you can decide, learn and implement S/BLOW support from this pull request if you like.
Note: slow5lib has been added as a submodule
so you must do a git submodule init and git submodule update and then make
alternatively recursively clone.
Is there a dataset that I can use to quickly verify if things are working as normal?
I ran the training on a simulated dataset (SARS-CoV-2 vs Yeast) , which gives a result like this:
##TRAINING MODE
#LOADING TARGET READS.........................
DONE loading 18831valid target reads
$$$$$$$$$$$$$$$$$$$$$$$$
#LOADING NON_TARGET READ......................
DONE loading 19501valid non-target reads
Trimmmed data size: 37662
OpenCV version 2.4.9.1 (2.4.9)
Using optimal parameters degree 0.000000, gamma 0.000010, ceof0 0.000000
C 0.100000, nu 0.000000, p 0.000000
Number of support vectors for trained SVM = 2000
Existing model file succesfully deleted
Model file created
As I had trouble getting the RawMap compiled with hdf5, I modified it to work with S/BLOW5. Had to comment out all FAST5-related stuff as I wanted to get it working fast, but you can decide, learn and implement S/BLOW support from this pull request if you like.
Note: slow5lib has been added as a submodule so you must do a
git submodule initandgit submodule updateand thenmakealternatively recursively clone.Is there a dataset that I can use to quickly verify if things are working as normal? I ran the training on a simulated dataset (SARS-CoV-2 vs Yeast) , which gives a result like this:
Not sure if it is as expected?