[Bioperl-l] Ensembl stable_id_event table

[Nathan (Nat) Goodman]

With apologies, this is not technically a bioperl question, but hopefully close enough.  I'm trying to map old Ensembl IDs to current ones.  Table stable_id_event on their ftp site seems to be the place to get this information, but the data in that table (for human build 61) is cyclic, meaning that old IDs are sometimes mapped to new ones which in turn are mapped back to the old one.  This seems wrong...  And, it's not a rare occurrence: 18% of IDs in the table lie on cycles, the largest of which contains more than 6000 IDs!

To see one short cycle, run this query (against human build 61).

mysql> select * from stable_id_event where 
    ->     (old_stable_id='ENSP00000399673' or new_stable_id='ENSP00000399673')
    -> and (old_stable_id='ENSP00000365536' or new_stable_id='ENSP00000365536');

The results are

+-----------------+-------------+-----------------+-------------+--------------------+-------------+-------+
| old_stable_id   | old_version | new_stable_id   | new_version | mapping_session_id | type        | score |
+-----------------+-------------+-----------------+-------------+--------------------+-------------+-------+
| ENSP00000399673 |           2 | ENSP00000365536 |           3 |                380 | translation |     1 | 
| ENSP00000399673 |           2 | ENSP00000365536 |           3 |                381 | translation |     1 | 
| ENSP00000365536 |           3 | ENSP00000399673 |           2 |                381 | translation |     1 | 
| ENSP00000399673 |           2 | ENSP00000365536 |           3 |                382 | translation |     1 | 
| ENSP00000365536 |           3 | ENSP00000399673 |           2 |                382 | translation |     1 | 
+-----------------+-------------+-----------------+-------------+--------------------+-------------+-------+

The first and last lines reveal the cycle:

line 1) ENSP00000399673-> ENSP00000365536
line 5) ENSP00000365536-> ENSP00000399673

If anyone can shed some light, I would be most grateful.

Thanks very much,
Nat Goodman