problem with overlapping/near identical attC sites

[jeanrjc]

observation: we may find 2 attC that are detected with the beginning of the 2nd attC that starts before the first attC ends:

ID_integron	ID_replicon	element	pos_beg	pos_end	strand	evalue	distance_2attC
integron_01	VIVU001.C.00013.C001	attc_014	1693083	1693209	-1.0	0.099	339
integron_01	VIVU001.C.00013.C001	attc_015	1693084	1693209	-1.0	0.017	3310967

Consequence: 2nd attC is one too many. The distance reported is thus wrong.

impact: minimal

We should remove duplicate for on pos_beg or pos_end here, not just pos_beg.

[bneron]

can you provide data to integrate them in tests?

[jeanrjc]

Hi @bneron, I think you have access to VIVU001.C.00013.C001, let me know if you can't reproduce the issue.

[bneron]

I haven't access to VIVU001.C.00013.C001

[bneron]

does the strand is important to determine duplicates?

[bneron]

if we remove attc sites which have pos_beg and pos_end close (less than 3 bases diff) what should the behavior if we have 3 attc with the following pos_beg/pos_end

Accession_number	cm_attC	cm_debut	cm_fin	pos_beg	pos_end	sens	evalue
ACBA.007.P01_13	attc_4	1	47	5500	7000	+	1.100000e-07
ACBA.007.P01_13	attc_4	1	47	5502	7002	+	1.100000e-04
ACBA.007.P01_13	attc_4	1	47	5504	7004	+	1.100000e-03

• The second attc is at less than 3 bases than the first one and should be removed ?
• The third attc is at less than 3 bases than the second one but more than 3 bases than the first one so should we remove or keep it ?

[cachapuz2001]

does the strand is important to determine duplicates?

I would say it is not. They are partly palindromic. I would suggest just check the overlap.

• The second attc is at less than 3 bases than the first one and should be removed ?
• The third attc is at less than 3 bases than the second one but more than 3 bases than the first one so should we remove or keep it ? My opinion: one should remove all that overlap more than 50% of the size. I.e. take the best one and remove those that overlap more than 50%