[Feedback] Site Feedback: 1-base shift in release-59/gff3/zea_mays/gene_id_mapping_v3_to_v4/V3gene.V4_coordinates.gff3

[feedback at gramene.org]

URL         : http://ensembl.gramene.org/Zea_mays/Info/Index

Subject     : 1-base shift in release-59/gff3/zea_mays/gene_id_mapping_v3_to_v4/V3gene.V4_coordinates.gff3

Name        : Wen-Dar Lin

Email       : wdlin at gate.sinica.edu.tw

Organization: IPMB, Academia Sinica

Comments    : I tried to work with v3 annotation on v4 genome so that I found the V3gene.V4_coordinates.gff3 file. An intreseting fact that I found is that the distribution of donor/accepter combination of this GFF3 file is not like of v3 annotation on v3 genome. I think GTAG should be the most common donor/accepter combination.

V3gene.V4_coordinates:
GGCA    56879
TAGG    32173
GTAG    24957
GGTA    15543
TAGA    13441
TGGG    10362
TAGC    7365
TAGT    6850
TGCA    6467
AGCA    5934
TTGG    5765
TGGA    4369
TCGG    3398
GGAA    3394
CGCA    2940
TTGA    2506
TGGC    2455
TGGT    2339
TGTA    1894
AGTA    1732

v3 annotation on v3 genome:
GTAG    259227
GCAG    4052
ATAC    189
CTAC    165
ATAG    150
GTAT    125
GTTG    121
GAAG    117
GTAA    115
GGAG    99
TTAG    95
GTGG    95
GTAC    93
CTAG    90
GTCG    75
GCCG    69
CCCG    61
ATAT    43
CTGC    39
CTGG    35

I built my version of v3-to-v4 GFF3 and I think I found the reason. Just an example: coordinates of a transcript from the V3gene.V4_coordinates.gff3
>GRMZM2G059865_T01
50882   51216
51370   51435
51885   52003
52136   52293
52390   52545
52667   52825
52946   53148
53622   53931
55221   55680

and mine:
GRMZM2G059865_T01
50883   51217
51371   51436
51886   52004
52137   52294
52391   52546
52668   52826
52947   53149
53623   53932
55222   55681

I understand that this might simply because of 0-base reports made by tools like MUMmer, but people used to consider coordinates in a GFF3 file as 1-base.

http://www.warelab.org/bugs/view.php?id=5784