screen.seqs similar sequences wildly different results

HI,
I was revisitng some of my old use cases and found this thing.
I had two sets of sequences which are rather similar by summary.seqs but when I apply similar screen.seqs procedures one of them gets trimmed to
pretty much nothing.
Set A
mothur > summary.seqs(fasta=MKB.trim.unique.align)
Using 7 processors.
Start End NBases Ambigs Polymer NumSeqs
Minimum: 1044 1048 2 0 1 1
2.5%-tile: 1044 1160 28 0 2 17513
25%-tile: 31137 32494 46 0 3 175121
Median: 31189 34113 71 0 3 350242
75%-tile: 32506 34113 76 0 4 525363
97.5%-tile: 42587 43103 80 0 5 682971
Maximum: 43115 43117 131 0 8 700483
Mean: 27861.2 29832.6 61.7814 0 3.50606

of Seqs: 700483

mothur > screen.seqs(fasta=MKB.trim.unique.align, name=MKB.trim.names, group=MKB.groups, end=34113,start=31189,minlength=65,processors=7)

mothur > summary.seqs(fasta=current)
Using MKB.trim.unique.good.align as input file for the fasta parameter.
Start End NBases Ambigs Polymer NumSeqs
Minimum: 28464 34113 65 0 2 1
2.5%-tile: 31187 34113 72 0 3 5555
25%-tile: 31189 34113 74 0 3 55550
Median: 31189 34113 76 0 4 111099
75%-tile: 31189 34113 77 0 4 166648
97.5%-tile: 31189 34113 82 0 5 216643
Maximum: 31189 35971 129 0 8 222197
Mean: 31187.2 34114 76.0732 0 3.6281

of Seqs: 222197

I can live with that.
Set B

mothur > summary.seqs(fasta=Ivika.trim.unique.align)
Start End NBases Ambigs Polymer NumSeqs
Minimum: 1044 1058 4 0 1 1
2.5%-tile: 1044 1162 31 0 2 11984
25%-tile: 31189 34113 53 0 3 119838
Median: 31189 34113 74 0 3 239676
75%-tile: 32485 34113 76 0 4 359513
97.5%-tile: 42584 43116 80 0 5 467367
Maximum: 43112 43117 139 0 8 479350
Mean: 28488.4 30608.2 65.369 0 3.56549

of Seqs: 479350

Start end etc. seems to be rahter similar
but after:
screen.seqs(fasta=Ivika.trim.unique.align, name=Ivika.trim.names, group=Ivika.groups, end=34113,start=31187,minlength=65,processors=8)

mothur > summary.seqs(fasta=current)
Using Ivika.trim.unique.good.align as input file for the fasta parameter.

Using 8 processors.
[WARNING]: This command can take a namefile and you did not provide one. The current namefile is Ivika.trim.good.names which seems to match Ivika.trim.unique.good.align.

Start End NBases Ambigs Polymer NumSeqs
Minimum: 28453 34113 70 0 2 1
2.5%-tile: 31124 34113 73 0 3 235
25%-tile: 31187 34113 75 0 3 2343
Median: 31187 34113 77 0 3 4685
75%-tile: 31187 34113 79 0 4 7027
97.5%-tile: 31187 34116 108 0 5 9134
Maximum: 31187 35461 139 0 8 9368
Mean: 31146.5 34116.9 79.2139 0 3.62126

of Seqs: 9368

Where did all the sequences go! What should I probably make even more forgiving in screen parameters for the second set?
PS.
BBCode table’s are not working in this forum? How are the table formatting guidelines for forum.
edit: the version of Mothur used for both was mothur v.1.28.0 64bit.

I’d encourage you to update to the most recent version of mothur, but that shouldn’t impact this problem.

I would probably remove the minlength parameter and just work with start and end. I guess datasets can vary a bit and that might be what you’re seeing here.

Pat