Dorado Demux Process Stuck After ~7000 Reads and Incorrect Summary File for --emit-fastq Option

[npanwar25]

Please describe the issue:
When running Dorado demux with default settings (without the --emit-fastq flag), the process gets stuck after processing approximately 7000 reads(Tried with multiple datasets). When using the --emit-fastq flag, the summary file incorrectly labels all demultiplexed reads as “unclassified.” The goal is to generate a correct summary file and ensure smooth processing of the raw FASTQ data without Dorado getting stuck.

Steps to reproduce the issue:
Run Dorado demux with the default settings (without the --emit-fastq flag) on a dataset of approximately 7000 reads.

Alternatively, run Dorado demux with the --emit-fastq flag and check the generated summary file, which incorrectly marks all reads as "unclassified." Issue is mentioned here: #872

Run environment:
Dorado version: dorado-0.8.1-osx-arm64
Operating system: macOS
Hardware: Mac OS, Apple M1 Pro, 10-core CPU, 32GB RAM, no GPU used.
Source data type: Raw FASTQ data (note: pod5 is recommended for optimal basecalling performance).
Source data location: Local device storage.
Details about data: Nanopore flow cell, R10.4.1 chemistry, using the DNA sequencing kit, approximately 400K reads.

[HalfPhoton]

Hi @npanwar25, apologies for the delay in responding.

Can you tell us the following:

1. Does this issue persist in 0.8.2?
2. How many reads are in the combined.fastq file. You say approximately 7000 reads - is it specifically 6899 as indicated by the log?
3. How did you create this fastq file?
4. Can you demux successfully with other files or does it always fail?

Best regards,
Rich