BAC library debugged

From Katia Wostrikoff and Olivier Vallon, IBPC, Paris:

While trying to complement mutations in specific regions of the genome using the BAC library from the Silflow/Lefebvre labs, we have come to realize that for many plates, the sequences indicated by the JGI browser (http://genome.jgi-psf.org/Chlre4/Chlre4.home.html) were not correct. We have systematically sequenced BAC ends from all plates of our copy of the library (obtained through the Clemson Genomics Institute, http://www.genome.clemson.edu/). We have found that for eleven of the forty 384-well plates, the sequence obtained from a particular well is actually that described by JGI and the old BAC converter (http://www.chlamy.org/bac_conv.html) as coming from the same well OF A DIFFERENT PLATE. We conclude that while rearraying the plates in 96-well format for sequencing, JGI probably mislabelled some of them, resulting in a mixup of accession numbers. For example, plate 21 would have been sequenced as if it were 20, 22 as 21 etc..

Our conclusions seem to be in line with the difficulties encountered by Jane Grimwood when trying to use specific BACs for genome finishing (personal communication), but they require independent confirmation. Please let us know (ovallon@ibpc.fr) if you have experienced similar problems. Also inform us immediately if in contrast you have used the incriminated plates without problems. This might indicate that we were mistaken, or that the BAC library distributed to different labs by Clemson are not identical…

In any event, this should not cast doubt on the quality of the work JGI has done in sequencing the genome. Mixups at sequencing do happen everywhere (for example, the first EST assembly at Stanford was affected by similar errors, see Jain et al (2007) Nucleic Acids Res. 35:2074). This does not affect the sequence assembly. It also does not affect the mapping data obtained with this library [Eukaryot Cell. 2003 Apr;2(2):362-79].

We are pleased to provide the community with a corrected Excel file that we believe now gives the exact correspondence between PTQ numbers and library well. We added a sheet allowing the reverse calculation (well to read), and a sheet listing the correspondence between the real and erroneous plate numbers. We also corrected small errors in the calculation, that led the old sheet to give the same well number for different PTQ accessions. Until the community confirms our conclusions, the old BAC converter will still be downloadable, but a cautionary note will be added.