spades --continue converting reads to binary every time

[milenovic]

Description of bug

When running spades (rnaspades in my case) to continue interrupted assembly, I see that it always first converts reads to binary, even though binary reads are already present in .bin_reads folder. Is this an expected behaviour? Can it be set to use the reads that are already converted?

The binary conversion takes a while and adds (seemingly) unnecessary time to the job.

The command used to start the initial job was rnaspades.py --checkpoints all --pe1-1 /mnt/lscratch/users/mmilenovic/data/WF-TWP-TWF/tTWF-TWP_1.fastq --pe1-2 /mnt/lscratch/users/mmilenovic/data/WF-TWP-TWF/tTWF-TWP_2.fastq --ss rf -t 28 -m 755 --phred-offset 33 -o /mnt/lscratch/users/mmilenovic/data/WF-TWP-TWF/WF-TWP-TWF_rDNA_rnaSPAdes_2

And to continue: rnaspades.py --continue -o /mnt/lscratch/users/mmilenovic/data/WF-TWP-TWF/WF-TWP-TWF_rDNA_rnaSPAdes_2

spades.log

spades.log

params.txt

params.txt

SPAdes version

SPAdes-3.15.3

Operating System

CentOS Linux release 7.9.2009

Python Version

Python 3.8.11

Method of SPAdes installation

manual compile from the source

No errors reported in spades.log

• [X] Yes

[olga24912]

Hi @milenovic! Thank you for your request!

You are looking for --no-clear-after option. You can use this option when you run spades.py first time. Something like this:

rnaspades.py --checkpoints all --no-clear-after --pe1-1 /mnt/lscratch/users/mmilenovic/data/WF-TWP-TWF/tTWF-TWP_1.fastq --pe1-2 /mnt/lscratch/users/mmilenovic/data/WF-TWP-TWF/tTWF-TWP_2.fastq --ss rf -t 28 -m 755 --phred-offset 33 -o /mnt/lscratch/users/mmilenovic/data/WF-TWP-TWF/WF-TWP-TWF_rDNA_rnaSPAdes_2

or if binary reads exist and you didn't write this option on the first run, you can run spades.py --restart-from last --no-clear-after instead of spades.py --continue

But if the file with bin_reads already lost the read conversion is necessary anyway.

I am worried that in this case the read conversion was only halfway through, and because of this, if you use this option, the results may be erroneous. Therefore, specifically for this case, I still recommend waiting for the conversion of the reads. But in the future you can use this option.

I would be glad if you try to use the --no-clear-after option and tell me if everything worked out.

Best, Olga