maxextend set to 100000 bp causes segfault

[jguhlin]

Thanks for this new tool. I'm currently exploring it, and am finding some very long repeats, hitting the upper maximum of maxextend (10kbp by default, so repeat hits 20kbp). Extending this to 20kbp gives me a few repeats of 40kbp in my genome, but at 100kbp the program segfaults with page memory. The genome is: Nomada fabriciana (nomad bee) from NCBI.

At these lengths, perhaps it is overkill for repeat biology, but if it's possible to resolve and then to break down manually, I'd be keen to try that.

The full command is

REPrise -input n_fabriciana_genomic_sanitized.fasta -output n_fabriciana_genomic_sanitized -pa 4 -maxextend 100000 -dist 1                                                                                                                                                          

Any advice?

Also, is it possible for the bed file to list the coordinates matching the repeat as well?

Cheers,
--Joseph

[atsuhmd]

Dear Joseph
Thanks for your comment.
I tried the same meaning of the command, but the program terminated successfully.

./REPrise -input ~/genome/NCBI/GCA_907165295.1_iyNomFabr1.1_genomic.fna -output test_d1 -maxextend 100000 -dist 1 -pa 8

The cause of the maximum repeat length is most likely due to tandem repeat, so it might be a good idea to use a genome with hard masked tandem repeat regions as input with software that masks tandem repeats (I used tantan in my paper).

The bed file is the region masked for repeat family detection. In other words, it is possible that the appropriate repeat family is found later than the previously masked region. Therefore, it is recommended to apply the reprof file to RepeatMasker for actual use as a repeat region. This is very confusing and will be fixed in the next version so that it is not output normally.

Regards,
Atsushi

[jguhlin]

Thanks! This helps a lot.

I'm getting an out of memory error now:

❯ REPrise -input tantan.fa -output e_adleriae_genomic_sanitized                                                                                                                                                                                                                                                                                                    (base)
NUMCHR 215946
Unable to allocate 323534193 bytes for the sequence

https://www.ncbi.nlm.nih.gov/datasets/taxonomy/1172129/

The assembly is mediocre quality, I'll try filtering out the smaller contigs:

sum = 314665288, n = 215946, ave = 1457.15, largest = 1261211
N50 = 38307, n = 1036
N60 = 10248, n = 2744
N70 = 4393, n = 7757
N80 = 2299, n = 17821
N90 = 1083, n = 37921
N100 = 20, n = 215946
N_count = 357660
Gaps = 737

It works after filtering out small contigs though (which is a good solution for me, maybe a README or a warning would be helpful)?

[atsuhmd]

Thanks for reporting the error.

REPrise has padding between input sequences to avoid detecting repeats across chromosomes.
However, it seems that the padding was consuming too much memory when a large number of contigs were input.
I will put it in the README. (I'd like to work on this in the future, but it will take some time to rewrite the alignment code.)

Thanks.

[jguhlin]

No problem, thanks for all your help with this.