Assembly size seems to double between contigger and polishing stage

[davidecarlson]

I'm assembling a somewhat low coverage (~20x) human genome using ONT R10.4.1 reads with flye v2.9.5-b1801.

During the contigger stage, the contigs.fasta file was ~2.9 GB in size. However, the final assembly.fasta file ended up being nearly double that size, ~5.8 GB.

I don't have any particular reason to think that the heterozygosity of this sample is especially high. Is it possible that alternative haplotypes were not collapsed?

In case it's useful, here is the command I ran:

flye \
--nano-hq ./ONT_data_combined.fastq.gz \
-o flye_out \
-t 96 \
--resume \
--scaffold

Thanks!
Dave

[mikolmogorov]

Hi Dave - never seen anything like this. I suspect this is some artifact with restart command - the only difference between contigs.fasta and assembly.fasta is polishing.

Could you try deleting 40-polishing directory and then use --resume-from polishing?

[davidecarlson]

Will do, thanks! I'll report back on the results.

[davidecarlson]

Happy to report that after restarting with --resume-from polishing, the final assembly size is 2.9 GB, just like the contigs.fasta file.

I'm not sure exactly what happened, so I'll let you close this or investigate further, whatever you prefer.

Thanks!
Dave

[mikolmogorov]

Thanks - I'll fix this in the next release!