Error: Unknown entries in loss dictionary:

[axgraf]

Hi,
I tried to run poreplex on a minION run using the follwing command to basecall and demultiplex barcodes from direct RNA sequencing:
poreplex -i 20191122_1000_MN24929_FAL01495_e9928ec8/fast5/ -o poreplex_demultiplex --barcoding --fast5 --basecall --parallel 20

This results in an error:

== Analysis settings ======================================
 * Input:	fast5/20191122_1000_MN24929_FAL01495_e9928ec8/fast5/	
 * Output:	poreplex_demultiplex
 * Processes:	20
 * Presets:	rna-r941.cfg
 * Basecall on-the-fly:		Yes (albacore 2.3.4)
 * Trim 3' adapter:		No
 * Filter concatenated read:	No
 * Separate by barcode:		Yes
 * Real-time alignment:		No
 * FASTQ in output:		Yes
 * FAST5 in output:		Yes
 * Basecall table in output:	No
===========================================================

==> Processing FAST5 files
ERROR: [signal_analyzer.py:48] Unhandled exception ValueError: Unknown entries in loss dictionary: ['config', 'class_name']. Only expected following keys: ['dense_2']Task exception was never retrieved
future: <Task finished coro=<ProcessingSession.run_process_batch() done, defined at /home/graf/.local/lib/python3.6/site-packages/poreplex/pipeline.py:193> exception=CancelledError()>
concurrent.futures._base.CancelledError

Is there anything I need to do to get around that error?

First I tried the Fast5 files basecalled with Guppy with the same result. So I thought running the internal albacore basecalling will help, but unfortunately no luck!
Could you give me a hint what to do to solve that issue?

Thank you in advance for your help
Best Alex

[chen318liang]

same problems! Poreplex is the latest version (v0.5). How to solve this bug??? it seems that many persons have counter this problem!

Poreplex version 0.5 by Hyeshik Chang hyeshik@snu.ac.kr

• Cuts nanopore direct RNA sequencing data into bite-size pieces for RNA Biology

Output directory poreplex_output/ is not empty. Clear it? (y/N) y

== Analysis settings ======================================

• Input: RS20200115_RNA/RS20200115_RNA_muliplex/RS20200115_RNA_multiplex/20200115_0837_MN23702_FAK99138_e3bc320c/fast5_result/workspace/
• Output: poreplex_output/
• Processes: 1
• Presets: rna-r941.cfg
• Basecall on-the-fly: No (use previous analyses)
• Trim 3' adapter: Yes
• Filter concatenated read: No
• Separate by barcode: Yes
• Real-time alignment: No
• FASTQ in output: Yes
• FAST5 in output: No
• Basecall table in output: No
  ===========================================================

==> Processing FAST5 files
ERROR: [signal_analyzer.py:48] Unhandled exception ValueError: Unknown entries in loss dictionary: ['class_name', 'config']. Only expected following keys: ['dense_2']WARNING: Logging before flag parsing goes to stderr.
E0117 14:43:27.784703 140625390106432 base_events.py:1285] Task exception was never retrieved
future: <Task finished coro=<ProcessingSession.run_process_batch() done, defined at /usr/local/lib/python3.6/dist-packages/poreplex/pipeline.py:193> exception=CancelledError()>
concurrent.futures._base.CancelledError
| 100% of 356000 |#######################################################################################################################################################################| Elapsed: 0:00:00 Time: 0:00:00

==> Terminated.

[lucapandolfini]

Which tensorflow version are you using? I suspect it is due to a tf version <2. In my case, using 2.0.0 fixed the issue.