Missing species designation when using sourmash lca classify on a Haemophilus influenzae

[replikation]

Dear sourmash team,

I have been using your great tool for years now and stumbled upon a strange behavior.

Issue: Missing species description when using sourmash lca classify on a Haemophilus influenzae genome.

Example fasta: https://eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=nuccore&id=1355929925&rettype=fasta
(I tried a few other H. influenza strains, but the same missing species issue. However, different species work fine, such as E.coli, S. aureus, etc.)
Database used: https://farm.cse.ucdavis.edu/~ctbrown/sourmash-db/gtdb-rs214/gtdb-rs214-k31.lca.json.gz
(same issue on an older DB)

version sourmash 4.8.14

Results:

<style> </style>

ID	status	superkingdom	phylum	class	order	family	genus	species	strain
NZ_CP020010.1 Haemophilus influenzae strain 67P38H1 chromosome, complete genome	found	d__Bacteria	p__Pseudomonadota	c__Gammaproteobacteria	o__Enterobacterales_A	f__Pasteurellaceae	g__Haemophilus

side info:
sourmash results when using sourmash gather

<style> </style>

intersect_bp	f_orig_query	f_match	f_unique_to_query	f_unique_weighted	average_abund	median_abund	std_abund	filename	name	md5	f_match_orig	unique_intersect_bp	gather_result_rank	remaining_bp	query_filename	query_name	query_md5	query_bp	ksize	moltype	scaled	query_n_hashes	query_abundance	query_containment_ani	match_containment_ani	average_containment_ani	max_containment_ani	potential_false_negative	n_unique_weighted_found	sum_weighted_found	total_weighted_hashes
1840000	1	1	1	1				gtdb-rs214-k31.lca.json.gz	GCF_002966675.1 Haemophilus influenzae strain=67P56H1, ASM296667v1	2f2a17a4bfe161d8e25ecbc0beaffd27	1	1840000	0	0	h.influenzae.fasta		e0f34f5b	184000	31	DNA	10000	184	FALSE	1	1	1	1	FALSE		184	184

[ctb]

thanks for reporting, @replikation!

I did:

grep GCF_002966675.1  gtdb-rs214.lineages.csv

and it does indeed have a species designation;

GCF_002966675.1,f,d__Bacteria,p__Pseudomonadota,c__Gammaproteobacteria,o__Enterobacterales_A,f__Pasteurellaceae,g__Haemophilus,s__Haemophilus influenzae

I am wondering if maybe the k-mers found were also largely shared with other species in g__Haemophilus? That would, I think, explain the result - LCA "pulls" the classification back to the lowest common ancestor that matches the hashes. Since you gave me enough info to reproduce, I'll take a look!

[replikation]

Great, let me know if you need anything else.