Cell-type tags in Protein and DNA sequence databases

Bob Phair rphair at ix.netcom.com
Sun Apr 6 11:10:48 EST 1997


While occasional records in dbEST do contain cell type information in
the LIBX field, the vast majority of new entries contributed by the
Washington Univ. - Merck consortium and by others are distinctly *not*
linked to specific cell types. Instead they often represent clones from
the normalized Soares libraries. If you examine the Library fields of
any of these records you will see something like:

"LIBRARY

Lib Name:       normalized infant brain cDNA
Organism:       Homo sapiens

Sex:            Female
Organ:          brain
Tissue type:    total brain
Develop. stage: 3 months old
Vector:         lafmid BA
R. Site 1:      HindIII
R. Site 2:      NotI
Description:    sex=Female; dev_stage=3 months old; isolate=muscular
atrophy
                patient; tissue_type=total brain; total mRNA was
oligo-(dT)
                primed and directionally cloned 5' -> 3' into the
HindIII ->
                NotI sites of the lafmid BA vector. Clone library from
B.Soares
,
                Psychiatry Dept. Columbia University, USA.
Normalization_method
:
                Bento Soares, P.N.A.S in press"

My original post was intended to highlight just this difficulty. A
normalized library from infant brain can never be used to assign an EST
to a particular cell type. "Total brain" will contain an untold number
of different cell types and the normalization process will have made
every mRNA from every cell type just as prominent as those from the cell
type you hope to study.

Another way to say this is to argue that cultured cells have many
advantages over whole tissue as the starting material for large scale
EST sequencing efforts. Cultured cells also have some disadvantages in
that the phenotype (and presumably gene expression) may change under
conditions of culture. But this might actually uncover genes that can be
expressed in a particular cell type, but that are not expressed in
normal circumstances. Differential display would also be valuable in
this context.

In summary, I remain concerned that the dbEST is being rapidly filled
with information that cannot be linked to mammalian cell biology and
physiology because the cell-types of origin are unknowable.

I'd value your further thoughts on this.

Regards,
Bob

--
Robert D. Phair, Ph.D.  rphair at ix.netcom.com
BioInformatics Services  http://www.webcom.com/rphair
Partnering and Outsourcing for Computational Biology





More information about the Comp-bio mailing list