Closed hkchi closed 4 years ago
hi there-- we are yet to update diploSHIC to be compatible with tensorflow 2+ as there have been major changes to the tf package. I'd recommend trying with tf 1.15 or similar.
Thanks for the reminder.
I tried both tensorflow 1.14.0
and tensorflow-gpu 1.14.0
with python 3.6.9
, and they worked well without changing outputFile.close
.
great! so if you are set i'll close this issue. feel free to reopen it if you are still having issues.
Hi,
Thanks for developing this useful tool and the detailed instructions coming with it.
I was testing diploSHIC using the Anopheles example, on a multicore machine using Intel E5-2683 v4 Broadwell CPU, with
python 3.7.6
,tensorflow 2.1.0
,keras 2.3.1
andpandas 1.0.1
from conda. By applying the proposed changes in #23 and #24, I was able to beat all the errors all the way down. However, the output file in predict mode was empty though it was generated. This can be fixed by changingoutputFile.close
tooutputFile.close()
at: https://github.com/kern-lab/diploSHIC/blob/7d55417af438e321c374dadc2f7e58f56107443e/diploSHIC.py#L319 Strangely, whenmodel_from_json
and the following prediction using the model were suppressed (no tensorflow invoked), contents fromoutputFile.write()
was visible in the output file without fixing theclose()
method.Kaichi