International Mammalian Genome Society

logo18th International Mouse Genome Conference

17-22 October 2004, Seattle, USA


ORAL PRESENTATION

MONDAY OCTOBER 18

3.00pm – 3.15pm

COUPLED COMPUTATIONAL AND EXPERIMENTAL APPROACHES TO DISCOVERY OF CODING AND NON-CODING GENES IN THE MOUSE GENOME

Hughes TR, Morris Q, Zhang W, Babak T, Mohammad N, Shai O, Fehlings M, Aubin J, van der Kooy D, Rossant J, Bruneau B, Blencowe B, Frey B

University of Toronto, Toronto, Canada

We are applying three different approaches to identify and characterize new transcripts in the mouse genome.  All three involve hybridizing RNA from diverse mouse tissues to custom Agilent oligonucleotide microarrays designed to detect computationally predicted genes. 

In the first approach, expression of 42,000 “XM” genes (NCBI predictions) was analyzed in 55 different tissues.  Using a stringent threshold, 21,575 genes were detected as “expressed”; among these, more than 5,000 are not present in current cDNA databases, and more than 3,000 are not present in EST databases.  Among the ~2,000 that were represented by an EST but not a cDNA, a random sampling demonstrated that more than half could be amplified by RT-PCR.  In virtually all cases we confirmed the tissue specificity observed on microarrays, which can be used to predict function.

In the second approach, we are using the program QRNA (Rivas and Eddy, 2001), which seeks regions of phylogenetically-conserved secondary structure, to predicting noncoding RNAs.  Using a pilot microarray to querying over 3,000 QRNA predictions across ten tissues, we detected fifty candidate noncoding RNAs as expressed.  Thus far we have confirmed three of these by Northern analysis that are between 50-100 nucleotides long and expressed at levels similar to the spliceosomal RNAs, but do not overtly appear to be members of any known RNA class (tRNA, snoRNA, etc.)

In the third approach, we designed microarrays containing over one million known and predicted mouse exons identified by one or more of five different gene-finding programs, and hybridized these arrays to cDNA from twelve mRNA pools encompassing forty different mouse tissues.  Visual analysis of the resulting data suggests that many sets of co-regulated exons are observed that appear likely to be novel, tissue-specific, multi-exon transcripts (i.e., they are not represented in RefSeq, Ensembl, Fantom, or Unigene).  Computational analysis of these data is in progress.

We anticipate that these analyses will contribute to the assembly of a complete collection of mammalian genes, and to their functional characterization.

[an error occurred while processing this directive]