About

[front page] [about] [programs] [databases] [complete genomes] [dna correlations] [mirror] 2008 | 2007 | 2006 | 2005 | 2004 | 2003 | 2002 | 2001 | 2000 | 1999 | 1998 | 1997 | 1996 | 1995 | 1994 | 1993 | 1992 | 1991 | 1990 | 1989 | 1988 | 1987 | 1986 | 1985 | 1984 | 1983 | 1982 | 1981 | 1980 | pre-1980 |

This web was created in 1996 to provide a collection of publications on computational gene recognition, mainly based on statistical differences of coding and non-coding sequences. With the complete sequence of many genomes available, including the human genome sequence, computational gene recognition becomes routine, and many non-statistical (i.e. biological) information is used for such a prediction. It is more and more difficult to decide what topics and what papers should be included, and can be of general interests. The first expansion of the topic was to include promoter and regulatory region recognition, which was already a deviation from the original plan to include only coding-noncoding recognition papers. A second expansion of the topic is to the use of microarray technology to monitor the expression of genes. A third expansion is gene annotation, which is a compilation of genes with their classifications and functions. And a fourth expansion is on alternative splicing.

Let's summary the original coding-noncoding distinction topic. How can such discrimination be made?

• We ask where the genes are unlikely to be located. [this is the strategy of excluding inter-genic regions]

• We ask how the transcription factors know where to bind a DNA region? [consensus patterns in promoter region, CpG islands, etc.]

• We ask how transcription, splicing, translation find their respective signals on the DNA sequence? [searching for the start codon, the stop codon, splicing sites, etc.] [be aware of the non-universal genetic code: ] http://www.ncbi.nlm.nih.gov/htbin-post/Taxonomy/wprintgc?mode=c]

• We ask what coding regions do "to make a living"? They make proteins! Since three nucleotides code for one amino acids, there is a distinct periodicity-three pattern in coding regions (but absent in noncoding regions). [this is the detection of periodicity-of-three signal strategy, and much of the earlier development in this field]

• We ask whether we can learn from examples, especially if these examples are obtained from the same species. [the codon usage strategy follows this direction]

• We ask whether we can "cheat" by looking up a "dictionary" (database of the known coding sequences). [this is sequence similarity search]

More challenges:

• The first and the last exon may contain untranslated regions (so-called 5' UTR and 3' UTR, or UTS where S stands for sequence) and the coding signal there can be very weak.

• Small exons are harder to predict.

• Since alternative splicing in human genome is not uncommon, we face the problem of non-unique solution of gene recognition.