duplex_tools filter_pairs failed on *fastq.gz files produced by Promethion about duplex-tools HOT 6 CLOSED

Thanks for spotting @ChuShin, we'll take a look at it and see what's going on

from duplex-tools.

Hm, curious, just to confirm, which version of duplex-tools was running there? I'm trying it on 0.2.5, and I think it should be functionally similar from 0.2.3 upwards. I think if something is going wrong with finding the expected ids, it would be here:

https://github.com/nanoporetech/duplex-tools/blob/master/duplex_tools/filter_pairs.py#L118

Is there any chance you could share the original fastq_pass/PAH93923_pass_4ac2ad82_13.fastq.gz from the INPUT section?

from duplex-tools.

Hi @ollenordesjo ,
I am using 0.2.5. Please find a link below to the test data:

https://www.dropbox.com/sh/wg6cj8q2hnag6hw/AABc_8kGEinYZ1h44s9PUyY7a?dl=0

Thanks

from duplex-tools.

Thanks @ChuShin! I'll just grab those files and have a go. Will keep you updated

from duplex-tools.

Hi @ChuShin, sorry for the delay in taking a look at this. I'm getting the one good pair as expected running with the test data you linked:

$ duplex_tools filter_pairs q30_test/pair_ids_test.txt q30_test/fastq_pass
[11:02:22 - FiltPairs] Filtering Parameters:
        nbases_per_strand_to_align=250
        align_threshold=0.6
[11:02:22 - FiltPairs] Alignment Parameters:
        score_match=2
        score_mismatch=-1
        penalty_open=4
        penalty_extend:1
[11:02:22 - ReadFastq] Processed 0/2 input fastq files.
[11:02:22 - ReadFastq] Found 100.0% of required reads.
[11:02:22 - FiltPairs] Starting alignments.
[11:02:22 - AlignPairs] Aligning 1 pairs
[11:02:22 - AlignPairs] Good pairs: 1

Would you mind trying by downloading the same data from the dropbox link you sent and see if that still gives an error for you? I'm wondering if this could somehow be platform-dependent.

Cheers!

from duplex-tools.

Hi @ollenordesjo ,

Thanks, you are right! I created a new virtual environment and got the expected results this time. The compatibility issue seems to have come from pip reusing existing libraries rather than collecting from duplex_tools at installation.

Cheers!

from duplex-tools.