Human-specific insertions and deletions inferred from mammalian genome sequences

Abstract

It has been suggested that insertions and deletions (indels) have contributed to the sequence divergence between the human and chimpanzee genomes more than do nucleotide changes (3% vs. 1.2%). However, although there have been studies of large indels between the two genomes, no systematic analysis of small indels (i.e., indels </= 100 bp) has been published. In this study, we first estimated that the false-positive rate of small indels inferred from human-chimpanzee pairwise sequence alignments is quite high, suggesting that the chimpanzee genome draft is not sufficiently accurate for our purpose. We have therefore inferred only human-specific indels using multiple sequence alignments of mammalian genomes. We identified >840,000 "small" indels, which affect >7000 UCSC-annotated human genes (>11,000 transcripts). These indels, however, amount to only approximately 0.21% sequence change in the human lineage for the regions compared, whereas in pseudogenes indels contribute to a sequence divergence of 1.40%, suggesting that most of the indels that occurred in genic regions have been eliminated. Functional analysis reveals that the genes whose coding exons have been affected by human-specific indels are enriched in transcription and translation regulatory activities but are underrepresented in catalytic and transporter activities, cellular and physiological processes, and extracellular region/matrix. This functional bias suggests that human-specific indels might have contributed to human unique traits by causing changes at the RNA and protein level.

Figure 1.

The analysis procedure flowchart.

Figure 2.

(A) Distributions of lengths and occurrences of human specific (HS) indels in different genomic regions. (B) Indel rate distributions of three kinds of indels: (1) indels retrieved from human–chimpanzee pairwise comparison, (2) indels retrieved from multiple sequence alignments (sequences from human, chimpanzee, and at least one species of mouse, rat, and dog), and (3) HS indels in different genomic regions. Note that category 3 indels are not considered here.

Figure 3.

Length distributions of human-specific (HS) indels (categories 1 and 2) in coding and noncoding regions.

Figure 4.

Gene ontology analysis of transcripts in which coding exons were affected by human-specific (HS) indels. The error bars indicate 95% confidence interval. The curves display comparisons of median K_a/K_s values for transcripts affected by CDS-HS indel(s) and transcripts without CDS-HS indel for each GO subcategory.

Figure 5.

Definitions of human-specific insertion (event A) and deletion (event B). Events C–F are not included in this study. In events C and D, the human specificity is uncertain. Events E and F represent nonhuman-specific indels.