The likelihood ratio test (LRT) of significantly conserved amino acid positions was applied to all condons within the human proteome. Codons with inadequate sequence alignments weren't  tested and are not listed in these files. The results are split to 343 files corresponding to non-overlapping 10 Mbp genomic blocks. The file “<A>.<B>.lrt” contains LRT values of all tested codons within the B-th 10 Mbp block in chromosome A. 

The .lrt files are tab-delimited plain text files that open with spreadsheet such as Excel or OpenOffice. Each line corresponds to each tested codon. The columns are:  

    Pos1, Pos2, Pos3: genomic coordinates of bases within the tested codon. Bases are ordered in reference to coding strand. 
    RefCodon: codon in the NCBI human reference genome.
    dS: synonymous substitution rate based on aligned species. Gaps in the alignment reduce dS of the tested site and the sensitivity of LRT. The maximum dS with all 31 vertebrates is 12.202.
    Omega: estimated nonsynonymous-to-synonymous-rate ratio.
    P: P-value of likelihood ratio test of significantly conserved codon.
    Alignment: amino acid alignment at tested codon. The '-' denotes a gap. The order of 31 aligned species is: 

        Chimpanzee, orangutan, rhesus, mouse lemur, bush baby, tree shrew, rat, mouse, squirrel,
        guinea pig, rabbit, pika, shrew, hedgehog, dog, cat, horse, cow, microbat, 
        armadillo, elephant, tenrec, opossum, platypus, chicken, frog, tetraodon, fugu,
        stickleback, medaka, zebra fish

At the beginning of each gene, a line starting with “#” character describes a gene name, a chromosome number, human Ensembl gene and peptide ID, and a coding strand (1 or -1) of the gene that follows. 

How to use this data to call deleterious nonsynonymous variants:
1. Locate the codon containing a tested variant. For instance, the variant at chrX:7033678 can be found at the line 7430 of the file "X.0.lrt". Information about the gene (HDHD1A) can be found at line 7342.
2. Call a variant “neutral” if P-value of LRT is above your choice of cutoff. The recommended cutoff is 0.001. 
3. Call a variant “neutral” if Omega is greater than 1, which indicates that the nonsynonymous rate is higher than the neutral rate. 
4. Call a variant “neutral” if the tested amino acid allele is observed in other placental mammals in the alignment. Otherwise, call “deleterious.” For instance, for a variant substituting G at chrX:7033678 to A, the variant amino acid allele (Met; A”T”G on the negative coding strand) is deleterious because it is not found in any placental mammals in the alignment. On the other hand, the reference amino acid allele (Thr; A”C”G on the negative coding strand) is neutral because it is found in placental mammals. 
5. Sometimes both reference and variant alleles are predicted deleterious. In this case, we recommend taking only derived allele as deleterious. We do not provide derived/ancestral allele assignment.

All coordinates in flat files are in the NCBI reference genome build 36, and gene predictions are from Ensembl release 49.

Contact schun@artsci.wustl.edu or jfay@genetics.wustl.edu for queries.
Please cite: 
Chun and Fay. Identification of deleterious mutations within three human genomes. Genome Research (2009) 19:1553-1561.

Last updated September 25, 2009.