AceView and the wonderful complexity of the human transcriptome.
Jean Thierry-Mieg 

AceView is an original engine which aligns all GenBank cDNAS, mRNAs and ESTs, to 
the genome, and clusters them into the minimal set of reconstructed transcripts 
compatible with all the data. Transcripts sharing one intron boundary are then 
considered alternative variants of a single gene. AceView and circumvents the 
sequencing noise in an auto adaptative fashion different from Blast and refines 
the exon boundaries by co-alignment. It does no mask repeats, does not 
precluster the cDNAs, does not use a priori splice consensus models, but groups 
reads from the same clone and carefully handles the cDNA cloning anomalies: we 
flag 5 to 10% of the clones as partial deletions, mosaics, strand inversions, 
internal priming or other rearrangements.
This allows us to use more experimental data than other programs, hence see the 
transcriptome in finer details. Our current release (build35/nov 04) assembles 
4.2 million human mRNAs and ESTs into over 40,000 genes putatively encoding more 
than 100 amino acids, and another 11,000 genes with standard introns but no 
clear protein coding potential. Alternative splicing is prevalent: the 33,000 
genes with standard introns have on average 5 mRNA variants per gene. In 
contrast to nematode, basically all spliced human genes have alternative forms. 
Our web site, www.ncbi.nlm.nih.gov/IEB/Research/Acembly, attempts to describe 
this complexity in detail. Web features that were fun to develop include an 
automatic text generator which has its adepts (1000 users a day),  a gene 
centered query language, a protein annotation pipeline. A lot more could be done 
and I will evoke the current potential uses as well as the future developments 
of the AceView/acedb database.