Genome Research (2003): Comparative Gene Prediction in Human and Mouse

SUPPLEMENTARY MATERIALS FOR


Comparative gene prediction

in human and mouse


G. Parra, P. Agarwal, J.F. Abril,

T. Wiehe, J.W. Fickett and R. Guigó ^*.


Genome Research 13(1):108-117 (Jan 1, 2003)

[ PubMed ]   [Abstract]   [Full Text]


* To whom correspondence should be adressed.

Email: rguigo@imim.es. Ph: +034 93-224-0877.

 

SGP2 is a program to predict genes by comparing anonymous genomic sequences from two different species. It combines tblastx (WU-Blast), a sequence similarity search program, with geneid, an ab initio gene prediction program. In assymetric mode, genes are predicted in one sequence from one species (the target sequence), using a set of sequences (maybe only one) from the other species (the reference set). Essentially, geneid is used to predict all potential exons along the target sequence. Scores of exons are computed as log-likelihood ratios, function of the splice sites defining the exon, the coding bias in composition of the exon sequence as measured by a Markov Model of order five, and of the optimal alignment at the amino acid level between the target exon sequence and the counterpart homologous sequence in the reference set. From the set of predicted exons, the gene structure is assembled (eventually multiple genes in both strands) maximizing the sum of the scores of the assembled exons.

CONTENTS

• SGP2: describing the algorithm.
   
• GFF2PS: how we made PostScript files for this page.
   
• SGP2 test sets
  • IMOG dataset.
  • BI dataset.
  • SCIMIT dataset.
     
• Gene predictions
  • 8 human/mouse homologous sequences.
    • Finished orthologous. Last Update: Aug 2, 2001
    • Finished human genome vs mouse reads. Last Update: Aug 17, 2001
       
  • Results on chromosome 22. Last Update: Oct 10, 2001
  • Whole genomes: Homo sapiens & Mus musculus. Last Update: May 6, 2002

SGP2 Test Sets

IMOG dataset

This is a list of 15 pairs of single gene sequences, with little overlap with the Sanger Center data set [Jareborg et al., Genome Research 9(9):815, 1999]. The gene accession-pairs associates a gene id to the corresponding human-mouse pair (first the human sequence, then the mouse).
 

• Gene accession-pairs list.
• Sequences in fasta format.
• Annotations in GFF format.

You can donwload a tarball containing all of above from here.

BI dataset

These are three pairs of multigene sequences. Annotation is not available for all the sequences, and of unknown reliability.
 

• Gene accession-pairs list.
• Sequences in fasta format.
• Annotations in GFF format.

You can donwload a tarball containing all of above from here.

SCIMIT dataset

This set contains 129 pairs of single gene sequences and combines non-overlaping sequences from IMOG (see above), the Sanger Center [Jareborg et al., Genome Research 9(9):815, 1999] and the MIT [Batzoglou et al., Genome Research 10(7):950, 2000] data sets. The gene accession-pairs associates a gene id to the corresponding human-mouse pair (first the human sequence, then the mouse).
 

• Gene accession-pairs list.
• Sequences in fasta format.
• Annotations in GFF format.

You can donwload a tarball grouping all of above from here.

Gene predictions: FINISHED HOMOLOGOUS SEQUENCES

Finished Orthologous

SGP2 predictions on the eigth human/mouse homologous sequences browsed from http://pipeline.lbl.gov/TESTS/ (including MHC). Unfortunately, that URL is no longer available. We just added a column (see Sequences and Annotations) to the table appearing below, which contains the human and mouse fasta files and the corresponding human annotations we have obtained from there.


Each human sequence was compared against the corresponding homologous mouse sequence.


We introduced few changes in the PostScript maps:
 

• We show the real length of annotated genes (taking into account first and last UTR coords), but we still display only the anontated CDS's.
• As we ran our programs on the original masked sequence, we are displaying only the masked regions for each sequence without labeling them (in the central axes of each block).
• We also included Twinscan results for the human/mouse homologous set.

Finished human vs. mouse reads

SGP2 predictions on the eigth human sequences browsed from http://pipeline.lbl.gov/TESTS/ (including MHC) against the mouse WGS 3X (a database of about 13 milion mouse reads). Unfortunately, that URL is no longer available, see previous table for the sequences and annotation of the ortologous datasets.


We used here human fasta sequences that were masked slightly different than those used in the human/mouse orthologous section.

Gene predictions: HUMAN CHROMOSOME 22

 

This section contains SGP2 predictions on human chromosome 22. Chromosome 22 annotation was compiled by Victoria Haghighi from the Columbia Genome Center. The data was downloaded from http://www.cs.columbia.edu/~vic/sanger2gbd.
There are two sets of SGP2 predictions. The first one are raw prediction along the whole Chromosome 22 sequence (Homology Only). The second one is a set of predictions confined to regions void of annotated genes or pseudogenes (Homology + Evidences). The goal is here predicting novel genes minimizing chimeric predictions. In this case, annotations are taken from the Combined Gene + CDS Set (879 genes).
 

Whole-Genome Gene-Predictions

 

The results of SGP2 on human and mouse genomes are available from our new Gene-Prediction section. Follow these links to download them: