trimLRPatterns with indels not trimming as intended in case of net insertions or deletions between subject and pattern?

[jan-glx]

In the following examples I'd expect trimLRPatterns to always leave ATCG:

library(Biostrings)
subject <- DNAStringSet(c("AAATCTCTCATCG"))
trimLRPatterns(subject = subject, Lpattern = "AAATCTCTC", Rpattern = "" , with.Lindels = TRUE, max.Lmismatch = 0.3 )
#> DNAStringSet object of length 1:
#>     width seq
#> [1]     4 ATCG
trimLRPatterns(subject = subject, Lpattern = "AAAATCTCTC", Rpattern = "" , with.Lindels = TRUE, max.Lmismatch = 0.3 )
#> DNAStringSet object of length 1:
#>     width seq
#> [1]     3 TCG
trimLRPatterns(subject = subject, Lpattern = "AATCTCTC", Rpattern = "" , with.Lindels = TRUE, max.Lmismatch = 0.3 )
#> DNAStringSet object of length 1:
#>     width seq
#> [1]     5 CATCG

^{Created on 2022-11-18 by the reprex package (v2.0.1)}

[acvill]

I think the issue is that Biostrings does not consider terminal gaps in alignments to be indels. See Bioconductor/pwalign#2.

ATCG is retained if you specify with.Lindels = FALSE.

subject <- DNAStringSet(c("AAATCTCTCATCG"))
trimLRPatterns(subject = subject, Lpattern = "AAATCTCTC", Rpattern = "" , with.Lindels = FALSE, max.Lmismatch = 0.3)
#> DNAStringSet object of length 1:
#>     width seq
#> [1]     4 ATCG
trimLRPatterns(subject = subject, Lpattern = "AAAATCTCTC", Rpattern = "" , with.Lindels = FALSE, max.Lmismatch = 0.3)
#> DNAStringSet object of length 1:
#>     width seq
#> [1]     4 ATCG
trimLRPatterns(subject = subject, Lpattern = "AATCTCTC", Rpattern = "" , with.Lindels = FALSE, max.Lmismatch = 0.3)
#> DNAStringSet object of length 1:
#>     width seq
#> [1]     6 TCATCG

[jan-glx]

Possibly. But while that issue might be considered odd behavior, this here is a bug, right?
Though by I am even more confused by your last with.Lindels = FALSE example right now.