Protein-coding region discovery in organisms underrepresented in databases

The prediction of coding sequences has received a lot of attention during the last decade. We can distinguish two kinds of methods, those that rely on training with sets of example and counter-example sequences, and those that exploit the intrinsic properties of the DNA sequences to be analyzed. The former are generally more powerful but their domains of application are limited by the availability of a training set. The latter avoid this drawback but can only be applied to sequences that are long enough to allow computation of the statistics. Here, we present a method that fills the gap between the two approaches. A learning step is applied using a set of sequences that are assumed to contain coding and non-coding regions, but with the boundaries of these regions unknown. A test step then uses the discriminant function obtained during the learning to predict coding regions in sequences from the same organism. The learning relies upon a correspondence analysis and prediction is presented on a graphical display. The method has been evaluated on a sample of yeast sequences, and the analysis of a set of expressed sequence tags from the Eucalyptus globulus-Pisolithus tinctorius ectomycorrhiza illustrates the relevance of the approach in its biological context.

Small genes/gene-products in Escherichia coli K-12

Forty-two protein spots of observed M(r) 6-15 kDa were resolved by two-dimensional gel electrophoresis, stained by Coomassie blue and subjected to Edman microsequencing. All of the proteins could be related back to their encoding open reading frames, thereby vindicating the bioinformatic tools currently utilised in their identification. However, only 14/42 gene-products were expressed as annotated. Translation was confirmed for 14 open reading frames with no attributed function (EcoGene Y-entries), while N-terminal sequence allowed the start codon to be accurately annotated for the genes yigF, yccU, yqiC, ynfD, and yeeX. The methionine start codon was cleaved in 11 gene-products (AtpE, Hns, RpoZ, RplL, CspC, YccJ, YggX, YjgF, HimA, InfA, RpsQ) and a further five showed loss of a signal peptide (PspE, HdeB, HdeA, YnfD, YkfE). Internal (Tig, AtpA, TufA) and N-terminal fragmentation (CspD, RpsF, AtcU) of much larger proteins was also detected, which may have resulted from physiological or translational processes. M(r) and pI isoforms were detected respectively for PtsH and GatB, each being phosphoproteins, as well as RplY which manifested differences with respect to predicted M(r) and pI. In addition, YjgF was shown to belong to a small gene family of unknown function with ancient conserved regions across procaryotes and eucaryotes. YgiN was revealed to have a paralogue and orthologues in Bacillus subtilis, Synechocystis sp., Mycobacterium tuberculosis, Neisseria gonorrhoea, and Rhodococcus erythropolis. Orthologues are also reported for YihD, YccU and YeeX. Of the 14 Y-genes, only YkfE possessed no detectable orthologues. These results highlight the need to complement genomic analysis with detailed proteomics in order to gain a better understanding of cellular molecular biology, while the confirmation of the open reading frame start codon using Edman degradation protein microsequencing has yet to be superseded by recent advances in mass spectrometry.

Characterization of new proteins found by analysis of short open reading frames from the full yeast genome

We have analysed short open reading frames (between 150 and 300 base pairs long) of the yeast genome (Saccharomyces cerevisiae) with a two-step strategy. The first step selects a candidate set of open reading frames from the DNA sequence based on statistical evaluation of DNA and protein sequence properties. The second step filters the candidate set by selecting open reading frames with high similarity to other known sequences (from any organism). As a result, we report ten new predicted proteins not present in the current sequence databases. These include a new alcohol dehydrogenase, a protein probably related to the cell cycle, as well as a homolog of the prokaryotic ribosomal protein L36 likely to be a mitochondrial ribosomal protein coded in the nuclear genome. We conclude that the analysis of short open reading frames leads to biologically interesting discoveries, even though the quantitative yield of new proteins is relatively low.