Issue with StrainScan_build of building a database using approximately 200 genomes

[zhangdan0]

Hello!

I am using approximately 200 refseq genomes to build a custom database, but program freezed and no progress. It seems to have stopped since yesterday, however process still exists. Here is the log output:

$ tail strainscan_build.o
Log: Clostridioides_difficile_strainCD161_GCF_003095675 179872 449680 5
Log: Clostridioides_difficile_strainDSM_29627_GCF_003482255 72064 72064 0
Log: Clostridioides_difficile_strainCDT4_GCF_003095695 74494 74494 0
Log: Clostridioides_difficile_strainSH_GCF_024972115 1122 1122 0
Log: Clostridioides_difficile_na_GCF_019095765 91342 91342 0
Log: Clostridioides_difficile_strainCD099_GCF_024972275 30548 30548 0
Unique part -> kid_match: 449442 , kmatrix: 449442
2025-05-19 18:06:11 - StrainScan::build_DB:: C13- Partial k-mer extract
Partial Unique part -> kid_match: 739960 , kmatrix: 739960
2025-05-19 18:06:17 - StrainScan::build_DB:: C13- Global k-mer extract and build matrix

I have tried to clear and rerun several times, but it doesn't work and get stuck in different clusters.

Looking forward to your guidance. Thanks!

[liaoherui]

Hi, apologies for the delayed response.

In theory, using 200 RefSeq genomes should be fine for building the database with the pipeline. One potential issue could be the similarity among these strains—if there's a very large cluster containing many similar strains, it might lead to high memory usage and long runtimes. You can check the file hclsMap_95_recls.txt to see the cluster sizes; clusters with more than 100 strains might be problematic.

If you don’t find any large clusters (e.g., cluster size >100), could you please share the genomes with me? I’d be happy to test them with the tool.

Sorry for the inconvenience, and thanks again!

[zhangdan0]

Thank you for your reply. Here is the file hclsMap_95_recls.txt.

hclsMap_95_recls.txt

Additionally, genomes list is uploaded together.

Clostridioides_difficile.txt

[liaoherui]

Really sorry for the late response — I’ve been tied up with ongoing projects. I’ve reviewed the genome list and don’t think the issue is due to sequence similarity, as I didn’t observe any large clusters.

I suspect the problem may be related to hardware limitations, such as insufficient memory. Unfortunately, I don't have time at the moment to download all the genomes and rebuild the database from scratch. However, if you’re able to provide the FASTA files directly, I’d be happy to help test the database building on my end.

Apologies for the inconvenience, and thank you for your patience and understanding!