[HELP] MacsyFinder not picking up MOB from CONJScan?

[milnus]

Describe your problem
I am trying to run MacsyFinder to classify/annotate conjugation machinery from plasmids.
When using MacsyFinder with the CONJScan model (v. 2.0.1) on multiple plasmids (~2000 using parallel_macsyfinder.nf) it resulted in no identified system, for any plasmid.

I then examined an example plasmid (attached).
Using blastp I found IMGPR_plasmid_2923296884_000005_2923296884_2923296888_4 to be a MOBQ relaxase.
I then tried to run MacsyFinder on the specific plasmid, to see if the problem was with parallel_macsyfinder.nf specifically, but still no MOB system was identified.
IMGPR_plasmid_2923296884_000005_2923296884_2923296888.faa.txt

Finally I inspected the hmmsearch outputs, that correctly identifies MOBQ as the best match for the IMGPR_plasmid_2923296884_000005_2923296884_2923296888_4 sequence.
T4SS_MOBQ.search_hmm.out.txt

Second best match (MOBP1 and MOBV)
T4SS_MOBP1.search_hmm.out.txt
T4SS_MOBV.search_hmm.out.txt

I do not know If I am missing something or there is a true problem at play. I focused on a MOB system, as it is simple and only requires a relaxase as a mandatory component for the system to be "called" (as I understand it).

To Reproduce
Steps to reproduce the behavior:

The exact command line used:

macsyfinder -o test_macsyfinder --db-type ordered_replicon --models CONJScan/Plasmids all --sequence-db IMGPR_plasmid_2923296884_000005_2923296884_2923296888.faa.txt

OS:

• Linux
• Windows
• Mac

MacSyFinder Version:

MacSyFinder 2.1.5
using:
- Python 3.13.7 | packaged by conda-forge | (main, Sep  3 2025, 14:24:46) [Clang 19.1.7 ]
- MacSyLib 1.0.3
- NetworkX 3.5
- Pandas 2.3.3

[saphia]

Hi!

The protein FASTA file you provide contain only a few proteins, is that your complete input for MacSyFinder (7 proteins)?
The rules of CONJscan request at least 3-4 (?) positive proteins to detect a potential conjugative system, falling under certain functional categories. These models do not really focus on annotating a single protein but rather look into the presence of what could be "complete" systems (also decayed ones in the particular case of CONJscan).

Also, maybe could you rather post an issue on the CONJscan page?
I'm not sure this is related to a MacSyFinder issue.

Best,

Sophie

[milnus]

Hi Saphia,

Yes, the Protein FASTA i linked only contains 7 genes. It is a very small plasmid, but was useful in examining my problem of the MOB system not being identified as this system only has a single mandatory gene component as in the rules of MOB outlines.

I will look a bit more into this and/or post on the CONJScan page.

Thanks
/Magnus

[bneron]

I have to dig little bit more but I think you found a bug.
The problem come from the building clusters phase, when I handle the circularity of plasmids.
it come when msf found only one cluster and the plasmid is small enough that the first hits and the last hits of this plasmid span the origin but close to colocates . I have to confirm that and find a solution.

[bneron]

I will publish a bug fix as soon as possible

[bneron]

May I use your input sequence with renaming genes to incorporate it in my regression tests?

[milnus]

May I use your input sequence with renaming genes to incorporate it in my regression tests?

Of course!