The European Functional Analysis Network (EUROFAN) is systematically analysing the function of novel Saccharomyces cerevisiae genes revealed by genome sequencing. As part of this effort our consortium has performed a detailed transcript analysis for 250 novel ORFs on chromosome XIV. All transcripts were quantified by Northern analysis under three quasi‐steady‐state conditions (exponential growth on rich fermentative, rich non‐fermentative, and minimal fermentative media) and eight transient conditions (glucose derepression, glucose upshift, stationary phase, nitrogen starvation, osmo‐stress, heat‐shock, and two control conditions). Transcripts were detected for 82% of the 250 ORFs, and only one ORF did not yield a transcript of the expected length (YNL285w). Transcripts ranged from low (62%), moderate (16%) to high abundance (2%) relative to the ACT1 mRNA. The levels of 73% of the 206 chromosome XIV transcripts detected fluctuated in response to the transient states tested. However, only a small number responded strongly to the transients: eight ORFs were induced upon glucose upshift; five were repressed by glucose; six were induced in response to nitrogen starvation; three were induced in stationary phase; five were induced by osmo‐stress; four were induced by heat‐shock. These data provide useful clues about the general function of these ORFs and add to our understanding of gene regulation on a genome‐wide basis. Copyright © 1999 John Wiley & Sons, Ltd.

upshift tend to employ favoured codons, whereas those overexpressed in starvation conditions do not. These results are interpreted in terms of a model in which competition between mRNA molecules for translational capacity selects for codons translated by abundant tRNAs. Keywords: gene expression/genome analysis/mRNA/ Saccharomyces cerevisiae/stress responses IntroductionThe availability of the complete genome sequence of the eukaryotic microorganism, Saccharomyces cerevisiae (Goffeau et al., 1996) has allowed researchers to monitor gene transcription on a global (or genome-wide) scale for the ®rst time. The resulting pro®les de®ne the complete set of mRNA molecules (the transcriptome; Velculescu et al., 1997) present in the yeast cell under a given set of physiological or developmental conditions (Oliver, 1997). Massively parallel analytical procedures are used in transcriptome analysis that involve the hybridization of labelled mRNA or cDNA molecules to arrays of`target' molecules representing all of the~6000 protein-encoding genes de®ned by the yeast genome (Mewes et al., 1997). These targets may be either oligonucleotides (Wodicka et al., 1997) or PCR products (Lashkari et al., 1997;Hauser et al., 1998) fabricated in either`micro' (on glass slides or chips; Lashkari et al., 1997;Wodicka et al., 1997) or`macro' (on nylon or polypropylene membranes; Hauser et al., 1998) formats. The mRNA or cDNA hybridization probes may be labelled either radioactively (usually with 33 P; Hauser et al., 1998) or¯uorescently (usually with Cy5 or Cy3; Winzeler et al., 1999). Whatever the experimental protocol employed, all transcriptome analyses using hybridization arrays have in common that they produce massive amounts of data that have to be`mined', using computational techniques, in order to extract meaningful biological information. A number of algorithms have been developed Brown et al., 2000;Kell and King, 2000) to permit the comparison of the transcription patterns of all 6000 protein-encoding genes in different physiological conditions or throughout a time course of development (Cho et al., 1998;Chu, 1998;Spellman et al., 1998) or physiological adaptation . While these algorithms are effective in clustering together genes that show similar patterns of regulation, it is clear that the composition of any particular cluster is enormously sensitive to the thresholds set either for transcript detection or for a signi®cant level of regulation, and thus to the way in which the data have been normalized or otherwise processed.Because of these concerns about data processing, it is important that we make use of existing biological knowledge in mining hybridization array data. This may be done in two ways, either empirically (e.g. by adjusting threshold levels until genes already known to be co-regulated are clustered together) or, more formally, by using supervised methods of machine learning (Brown et al., 2000;Kell and King, 2000). Whatever approach is used, there is the problem that the prior knowledge has been gained using a di...

No abstract

We report the sequence analysis of a 10 kb DNA fragment of Saccharomyces cerevisiae chromosome VII. This sequence contains four complete open reading frames (ORFs) of greater than 100 amino acids. There are also two incomplete ORFs flanking the extremes: one of these, G2868, is the 5′ part of the SCS3 gene (Hosaka et al., 1994). ORFs G2853 and G2856 correspond to the genes CEG1, coding for the alfa subunit of the mRNA guanylyl transferase and a 3′ gene of unknown function previously sequenced (Shibagaki et al., 1992). G2864 is identical to SOH1 also reported (Fan and Klein, 1994). The translated sequence of G2861 is similar to the human dnaJ homolog. The nucleotide sequence reported here has been entered in the EMBL Data Library under the Accession Number X87252.