Using 1000 genomes data files

The 1000 genomes project was established in 2008, a few years after the human genome project (HGP) was completed in 2001. While the HGP was aimed at producing a single reference genome for Homo sapiens, the goal of the 1000 genomes project was to identify the variation present within the species by sequencing 1000s of individuals using fast and cheaper next generation sequencing (NGS) technology. Both projects put their data on line for researchers to download for free.

FASTA Files

The HGP packages the human genome into FASTA format files with one file for every human chromosome (1..22, X, Y) and an accessory FASTA file for the mitochondrial DNA (mtDNA). The FASTA format is a really simple text file format and its suffix tends to end in .fa, .fas or .fasta. The first line of the file consists of a descriptive header beginning with the greater-than symbol '>' followed by the name and/or description and/or unique identifier of the sequence. The next lines after that is the sequence. Typically, 60 characters are present on a line, followed by a new line character '\n', and this pattern goes on till the end of the sequence. Here is an example of part of the Human Cytochrome C gene:

>chromosome:GRCh38:7:25119491:25125961:-1 CTGTCTCTGCCGCAAATGCAGCACCTTCCTCAGTCTTGGGGCGTCCATCAAGCGGAGAGC TGGAATGTGACTGAAAGTAACAGAGTATAGGTGGAACTAGAACAAGGTCACGAGCTGGTG CCCAACTCGTGTAACTCCTGGTTCCAGGCTGTCCCTCTAGGATGCTGTTGGCGCCAGCCG TGGTAGTAGGACAGAGTCTGGGTAAAGGTCGCCTGCTGACAAGATGGTAGCTAGGAGACC TCTAGGTGCGCCAAAATTAGGGCGTCTTTTCC...(etc)

As you can see, the first line describes where it is obtained from. GRCh38 is the latest version of the HGP reference sequence. The gene can be found on chromosome 7 and starts from position 25119491 and ends at position 25125961. -1 indicates it is encoded on the reverse strand. The lines after are the actual genetic code.

The HGP package each assembled whole chromosome's worth of genetic data into individual FASTA files of between 50 and 250MB in size. You can download the gzip-compressed files by going to the ENSEMBL FTP site: ftp://ftp.ensembl.org/pub/release-86/fasta/homo_sapiens/dna/

An alternative to downloading each chromosome from the ENSEMBL site is to download an 860MB compressed file containing all the chromosomes in one big FASTA file from the 1000 genomes project. In this blog post I suggest downloading the slightly older GRCh37 version of the HGP, specifically file hs37d5.gz, because it is fully compatible with the VCF files mentioned below. You can obtain the reference genome from here: ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/technical/reference/phase2_reference_assembly_sequence/

You don't have to decompress that gigantic FASTA file because I wrote a little utility application to extract specific chromosomes from the large gzip file and output the chromosome in its own FASTA file: https://github.com/webmasterar/extractChromosome


After you download a chromosome file from ENSEMBL you can just decompress it using your favourite decompression utility such as winrar or if you are on Linux you can write the command: gzip -d Homo_sapiens.GRCh38.dna_rm.chromosome.21.fa.gz.

FASTA is a simple and flexible format. There are no restrictions on the number of sequences in a file as long as they each start with a header line followed by the sequence. Such files are typically known as multi-FASTA files, but they don't have a specific file suffix to indicate that. The format allows for comment lines starting with a semi-colon ';'. There are no restrictions on line length either and a whole sequence could be present on a single line. Typically, when reading FASTA files in a program, they are read line by line, stripping line-ending characters and spaces, skipping past empty and comment lines starting with ';', and storing the title and sequence in memory as strings ready to be searched or analysed.

VCF Files

VCF files are also text files but are a more complex format than FASTA. VCF files are used to store variants, typically single nucleotide polymorphisms (SNPs) and consist of a set of mandatory and optional tab-separated columns. They can represent more complex variants consisting of insertions and deletions too. VCF files are used together with FASTA reference chromosome files and would be of little use by themselves. Each row of the VCF file is a record of a SNP/variant and points to a position in the reference chromosome where the variation occurs. So together, it is possible to map all variants from multiple donors onto the reference human genome.

VCF files are often distributed as specially compressed files that might resemble gzip files (since they end with .gz) but are actually compressed using BGZIP. The compression utility BGZIP is part of a package known as Tabix, both of which are now part of the htslib library and packaged with SAMTools. If you ever need to use BGZIP you need only download the Tabix package using your Linux distro's package manager, e.g. sudo apt-get install tabix. Tabix contains a library for reading tab-separated file formats and together with BGZIP it allows applications to read the tab-separated VCF format in its compressed state. This is especially important because VCF files are huge. Chromosome 1's compressed VCF file is 1.2GB in size and when it is decompressed it takes up around 70GB of disk space! Imagine having to decompress all the human chromosomes - a time-consuming and disk-consuming exercise that is ultimately unnecessary due to Tabix and its BGZIP compression algorithm.

The VCF format contains rows separated by lines and each line is separated by tabs for each of the columns. The first few lines in the file are metadata and comments starting with a hash '#' character and contain details about the file and it's corresponding reference sequence, and information about indels and other complex variants as well as details about optional column headers. The mandatory headers are:

#CHROM - the chromosome number
POS - the position that the SNP/variant occurs in the reference sequence
ID - just the ID of the SNP/variant
REF - the base that should be found in the reference sequence at this position
ALT - the alternate base or bases that might be present here*
QUAL - the quality of the read
FILTER - only accept this record if it says 'PASS'
INFO - various details about phasing and such

*The ALT field may contain references to complex variants written in angle brackets like so '<ADD:INS:1>'.

Here is an example record from a VCF file:

#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT  HG00096
10   60523  rs148087467    T     G       100     PASS    AC=0;AF=0.01;AFR_AF=0.06;AMR_AF=0.0028;AN=2; GT:GL 0|0:-0.19,-0.46,-2.28

You can see it is a T to G single nucleotide polymorphism occurring at position 60523 on chromosome 10.

Most VCF files list the variants in ascending order because many software applications that read VCF files expect it to be in order. It is possible to have multiple entries for the same position of a chromosome in VCF files and thus have duplicates. Dealing with this issue will be resolved in a future blog post. Typically, you would not write your own parser to go through the VCF file because it has a complex compression and file format. For a C++ project I used VCFLib.

The 1000 genomes project was performed in three major phases. The first or pilot phase obtained DNA from 179 individuals, the second (Phase 1) from 1092 individuals (including the pilots) and the third phase (Phase 3) from 2504 individuals (including the people from the first two phases). They were sequenced with improving NGS technologies and the data was collected into VCF files for each chromosome. Since this project started in 2008, the reference human genome at the time was GRCh37 and as such the VCF files should use that as it's reference format. This is mentioned in the comments of the phase 3 VCF files. When you download the compressed (.gz) VCF file you will also need to download its accompanying Tabix index file (.tbi) which contains the index that helps Tabix read the file in its compressed state. You DO NOT need to decompress the VCF files. The files for phase 3 can be downloaded from here: ftp://ftp.1000genomes.ebi.ac.uk/vol1/ftp/release/20130502/

In conclusion, to use the 1000 genomes files, you will need to download the reference human genome sequence chromosome by chromosome (or as one big 850MB compressed file), the VCF file for the chromosome and its accompanying TBI file. There are is no need to decompress the files.

10 years of the Human Genome

The completion of the draft human genome sequence was announced ten years ago. Nature 's survey of life scientists reveals that biology will never be the same again. Declan Butler reports.

Declan Butler

Download a PDF of this story.

"With this profound new knowledge, humankind is on the verge of gaining immense, new power to heal. It will revolutionize the diagnosis, prevention and treatment of most, if not all, human diseases." So declared then US President Bill Clinton in the East Room of the White House on 26 June 2000, at an event held to hail the completion of the first draft assemblies of the human genome sequence by two fierce rivals, the publicly funded international Human Genome Project and its private-sector competitor Celera Genomics of Rockville, Maryland (see Nature 405, 983–984; 2000).

Ten years on, the hoped-for revolution against human disease has not arrived — and Nature 's poll of more than 1,000 life scientists shows that most don't anticipate that it will for decades to come (go.nature.com/3Ayuwn). What the sequence has brought about, however, is a revolution in biology. It has transformed the professional lives of scientists, inspiring them to tackle new biological problems and throwing up some acute new challenges along the way.

Almost all biologists surveyed have been influenced in some way by the availability of the human genome sequence. A whopping 69% of those who responded to Nature 's poll say that the human genome projects inspired them either to become a scientist or to change the direction of their research. Some 90% say that their own research has benefited from the sequencing of human genomes — with 46% saying that it has done so "significantly". And almost one-third use the sequence "almost daily" in their research. "For young researchers like me it's hard to imagine how biologists managed without it," wrote one scientist.

“69% were inspired by the genome to become a scientist or change their research direction.”

The survey, which drew most participants through Nature 's print edition and website and was intended as a rough measure of opinion, also revealed how researchers are confronting the increasing availability of information about their own genomes. Some 15% of respondents say that they have taken a genetic test in a medical setting, and almost one in ten has used a direct-to-consumer genetic testing service. When asked what they would sequence if they could sequence anything, many respondents listed their own genomes, their children's or those of other members of their family (the list also included a few pet dogs and cats).
Some are clearly impatient for this opportunity: about 13% say that they have already sequenced and analysed part of their own DNA. One in five said they would have their entire genome sequenced if it cost US$1,000, and about 60% would do it for $100 or if the service were offered free. Others are far more circumspect about sequencing their genome — about 17% ticked the box saying "I wouldn't do it even if someone paid me".

Click for a larger version.

Nature 's poll also gauged where the sequence has had the greatest effect on the science itself. Although nearly 60% of those polled said they thought that basic biological science had benefited significantly from human genome sequences, only about 20% felt the same was true for clinical medicine. And our respondents acknowledged that interpreting the sequence is proving to be a far greater challenge than deciphering it. About one-third of respondents listed the field's lack of basic understanding of genome biology as one of the main obstacles to making use of sequence data today.

Sequence is just the start

Studies over the past decade have revealed that the complexity of the genome, and indeed almost every aspect of human biology, is far greater than was previously thought (see Nature 464, 664–667; 2010). It has been relatively straightforward, for example, to identify the 20,000 or so protein-coding genes, which make up around 1.5% of the genome. But knowing this, researchers note, does not necessarily explain what those genes do, given that many genes code for multiple forms of a protein, each of which could have a different role in a variety of biological processes. "The total sequence was needed, I think, to allow us to see that our one gene–one protein model of genetics was much too simplistic," wrote one respondent.

A decade of post-genomic biology has also focused new attention on the regions outside protein-coding genes, many of which are likely to have key functions, through regulating the expression of protein-coding genes and by making a slew of non-coding RNA molecules. "Now we understand," wrote another survey respondent, "that, without looking at the dynamics of a genome, determining its sequence is of limited use." Some big projects are under way to fill in the gaps, including the Encyclopedia of DNA Elements (ENCODE) and the Human Epigenome Project, an effort to understand the chemical modifications of the genome that are now thought to be a major means of controlling gene expression.

The biggest effects of the genome sequence, according to the poll, have been advances in the tools of the trade: sequencing technologies and computational biology. Technological innovation has sent the cost of sequencing tumbling, and the daily output of sequence has soared (see Nature 464, 670–671; 2010). "Deep sequencing technology is now becoming a staple of scientific research. Would this have occurred if it wasn't for the technological push required to finish the human genome?" read one response.

Data dreams, analysis nightmares

Cheaper and faster sequencing has brought its own problems, however, and our survey revealed how ill-equipped many researchers feel to handle the exponentially increasing amounts of sequence data. The top concern — named by almost half of respondents — was the lack of adequate software or algorithms to analyse genomic data, followed closely by a shortage of qualified bioinformaticians and to a lesser extent raw computing power. Other concerns include data storage, the quality of sequencing data and the accuracy of genome assembly. Commenting on the survey results, David Lipman, director of the US National Center for Biotechnology Information in Bethesda, Maryland, says that the worries about data handling and analysis were an issue even in the earliest discussions of the genome project. Perhaps, he suggests, "there's a sort of disappointment that despite having so much data, there is still so much we don't understand".

Click for a larger version.

Eric Green, director of the National Human Genome Research Institute (NHGRI) in Bethesda, says that the institute is well aware of the need for more bioinformatics experts, better software and a clearer understanding of how the differences between genomes influence human health. He says the institute is planning to publish in late 2010 its next strategic five-year plan for the genomics field. One possible solution to the computing challenge, which was discussed at an NHGRI workshop in late March, is cloud computing, in which laboratories buy computing power and storage in remote computing farms from companies such as Google, Amazon and Microsoft. The European Nucleotide Archive, launched on 10 May at the European Molecular Biology Laboratory's European Bioinformatics Institute in Cambridge, UK, will also offer labs free remote storage of their genome data and use of bioinformatics tools.

“13% have sequenced part of their own DNA.”

Given ten years' of hindsight and the current set of obstacles, it's no surprise that researchers now state somewhat modest expectations for what human genomics can deliver and by when. The rationale for sequencing and exploring the human genome — to revolutionize the finding of new drugs, diagnostics and vaccines, and to tailor treatments to the genetic make-up of individuals — is the same today. But almost half of respondents now say that the benefits of the human genome were oversold in the lead up to 2000. "While I do feel that the gains made by the human genome project are extraordinary and affect my research significantly, I still feel that it was overhyped to the general population," read one typical response. More than one-third of respondents now predict that it will take 10–20 years for personalized medicine, based on genetic information, to become commonplace, and more than 25% even longer than that. Some 5% don't expect it will happen in their lifetime. "Our understanding of the genome will not come in a single flash of insight. It will be an organized hierarchy of billions of smaller insights," says David Haussler, head of the Genome Bioinformatics Group at the University of California, Santa Cruz.

Green says that when the Human Genome Project was envisioned, scientific leaders of the day predicted that it would take 15 years to generate the first sequence, and a century for biologists to understand it. "I think they got that about right," he says. "While we still don't have all the answers — being a mere 10% of the way into the century with a human genome sequence in hand — we have learned extraordinary things about how the human genome works and how alterations in it confer risk for disease."
Haussler agrees. "All that happened in the first ten years is still just early rumblings of much more dramatic changes to come when we begin to truly understand the genome," he says.

http://www.nature.com/news/2010/100623/full/4651000a.html

I, virus: Why you're only half human

WHEN, in 2001, the human genome was sequenced for the first time, we were confronted by several surprises. One was the sheer lack of genes: where we had anticipated perhaps 100,000 there were actually as few as 20,000. A bigger surprise came from analysis of the genetic sequences, which revealed that these genes made up a mere 1.5 per cent of the genome. This is dwarfed by DNA deriving from viruses, which amounts to roughly 9 per cent.

On top of that, huge chunks of the genome are made up of mysterious virus-like entities called retrotransposons, pieces of selfish DNA that appear to serve no function other than to make copies of themselves. These account for no less than 34 per cent of our genome.

All in all, the virus-like components of the human genome amount to almost half of our DNA. This would once have been dismissed as mere "junk DNA", but we now know that some of it plays a critical role in our biology. As to the origins and function of the rest, we simply do not know.

The human genome therefore presents us with a paradox. How does this viral DNA come to be there? What role has it played in our evolution, and what is it doing to our physiology? To answer these questions we need to deconstruct the origins of the human genome - a story more fantastic than anything we previously imagined, with viruses playing a bigger part than you might care to believe.

Around 15 years ago, when I was researching my book Virus X, I came to the conclusion there was more to viruses than meets the eye. Viruses are often associated with plagues - epidemics accompanied by great mortality, such as smallpox, flu and AIDS. I proposed that plague viruses also interact with their hosts in a more subtle way, through symbiosis, with important implications for the evolution of their hosts. Today we have growing evidence that this is true (New Scientist, 30 August 2008, p 38), and overwhelming evidence that viruses have significantly changed human evolution.

Symbiosis was defined by botanist Anton de Bary in 1878 as the living together of dissimilar organisms. The partners are known as symbionts and the sum of the partnership as the holobiont. Types of symbiotic relationships include parasitism, where one partner benefits at the expense of the other, commensalism, where one partner profits without harming the other, and mutualism, in which both partners benefit.

Symbiotic relationships have evolutionary implications for the holobiont. Although selection still operates on the symbionts at an individual level since they reproduce independently, it also operates at partnership level. This is most clearly seen in the pollination mutualisms involving hummingbirds and flowers, where the structure of flower and bill have co-evolved to accommodate each other and make a perfect fit. When symbiosis results in such evolutionary change it is known as symbiogenesis.

Viruses as partners

Symbiosis works at many different levels of biological organisation. At one end of the spectrum is the simple exchange of metabolites. Mycorrhizal partnerships between plant roots and fungi, which supply the plant with minerals and the fungus with sugars, are a good example. At the other end are behavioural symbioses typified by cleaning stations where marine predators line up to have their mouths cleared of parasites and debris by fish and shrimps.

Symbiosis can also operate at the genetic level, with partners sharing genes. A good example is the solar-powered sea slug Elysia chlorotica, which extracts chloroplasts from the alga it eats and transfers them to cells in its gut where they supply the slug with nutrients. The slug's genome also contains genes transferred from the alga, without which the chloroplasts could not function. The slug genome can therefore be seen as a holobiont of slug genes and algal genes.

This concept of genetic symbiosis is crucial to answering our question about the origin of the human genome, because it also applies to viruses and their hosts. Viruses are obligate parasites. They can only reproduce within the cells of their host, so their life cycle involves forming an intimate partnership. Thus, according to de Bary's definition, virus-host interactions are symbiotic.

Genetic symbiosis is crucial to understanding the origin of the human genome, because it also applies to viruses

For many viruses, such as influenza, this relationship is parasitic and temporary. But some cause persistent infections, with the virus never leaving the host. Such a long-term association changes the nature of the symbiosis, making the evolution of mutualism likely. This process often follows a recognisable progression I have termed "aggressive symbiosis".

An example of aggressive symbiosis is the myxomatosis epidemic in rabbits in Australia in the 1950s. The European rabbit was introduced into Australia in 1859 as a source of food. Lacking natural predators, the population exploded, leading to widespread destruction of agricultural grassland. In 1950, rabbits infected with myxoma virus were deliberately released into the wild. Within three months, 99.8 per cent of the rabbits of south-east Australia were dead.

In 1950, rabbits infected with myxoma virus were released into the wild. Within three months 99.8 per cent of rabbits in south-east Australia were dead

Although the myxomatosis epidemic was not planned as an evolutionary experiment, it had evolutionary consequences. The myxoma virus's natural host is the Brazilian rabbit, in which it is a persistant partner causing no more than minor skin blemishes. The same is now true of rabbits in Australia. Over the course of the epidemic the virus selected for rabbits with a minority genetic variant capable of surviving infection. Plague culling was followed by co-evolution, and today rabbit and virus coexist in a largely non-pathogenic mutualism.

Now imagine a plague virus attacking an early human population in Africa. The epidemic would have followed a similar trajectory, with plague culling followed by a period in which survivors and virus co-evolved. There is evidence that this happened repeatedly during our evolution, though when, and through what infectious agents, is unknown (Proceedings of the National Academy of Sciences, vol 99, p 11748).

Even today viral diseases are changing the course of human evolution. Although the plague culling effect is mitigated by medical intervention in the AIDS pandemic, we nevertheless observe selection pressure on humans and virus alike. For example, the human gene HLA-B plays an important role in the response to HIV-1 infection, and different variants are strongly associated with the rate of AIDS progression. It is therefore likely that different HLA-B alleles impose selection pressure on HIV-1, while HLA-B gene frequencies in the population are likely to be influenced by HIV (Nature, vol 432, p 769). This is symbiogenesis in action.

How does that move us closer to understanding the composition of the human genome? HIV-1 is a retrovirus, a class of RNA virus that converts its RNA genome into DNA before implanting it into host chromosomes. This process, known as endogenisation, converts an infectious virus into a non-infectious endogenous retrovirus (ERV). In humans, ERVs are called HERVs.

Germline invaders

Endogenisation allows retroviruses to take genetic symbiosis to a new level. Usually it is an extension of the normal infectious process, when a retrovirus infects a blood cell, such as a lymphocyte. But if the virus happens to get incorporated in a chromosome in the host's germ line (sperm or egg), it can become part of the genome of future generations.

Such germ-line endogenisation has happened repeatedly in our own lineage - it is the source of all that viral DNA in our genome. The human genome contains thousands of HERVs from between 30 and 50 different families, believed to be the legacy of epidemics throughout our evolutionary history. We might pause to consider that we are the descendents of the survivors of a harrowing, if brutally creative, series of viral epidemics.

Endogenisation is happening right now in a retroviral epidemic that is spreading among koalas in Australia. The retrovirus, KoRv, appeared about 100 years ago and has already spread through 75 per cent of the koala's range, culling animals on a large scale and simultaneously invading the germ line of the survivors.

Retroviruses don't have a monopoly on endogenisation. Earlier this month researchers reported finding genes from a bornavirus in the genomes of several mammals, including humans, the first time a virus not in the retrovirus class has been identified in an animal genome. The virus appears to have entered the germ line of a mammalian ancestor around 40 million years ago (Nature, vol 463, p 84). Many more such discoveries are anticipated, perhaps explaining the origin of some of that mysterious half of the genome.

The ability of viruses to unite, genome-to-genome, with their hosts has clear evolutionary significance. For the host, it means new material for evolution. If a virus happens to introduce a useful gene, natural selection will act on it and, like a beneficial new mutation, it may spread through the population.

Could a viral gene really be useful to a mammal? Don't bet against it. Retroviruses have undergone a long co-evolutionary relationship with their hosts, during which they have evolved the ability to manipulate host defences for their own ends. So we might expect the genes of viruses infecting humans to be compatible with human biology.

This is also true of their regulatory DNA. A virus integrating itself into the germ line brings not just its own genes, but also regulatory regions that control those genes. Viral genomes are bookended by regions known as long terminal repeats (LTRs), which contain an array of sequences capable of controlling not just viral genes but host ones as well. Many LTRs contain attachment sites for host hormones, for example, which probably evolved to allow the virus to manipulate host defences.

Retroviruses will often endogenise repeatedly throughout the host genome, leading to a gradual accumulation of anything up to 1000 ERVs. Each integration offers the potential of symbiogenetic evolution.

Once an ERV is established in the genome, natural selection will act on it, weeding out viral genes or regulatory sequences that impair survival of the host, ignoring those that have no effect, and positively selecting the rare ones that enhance survival.

Most ERV integrations will be negative or have no effect. The human genome is littered with the decayed remnants of such integrations, often reduced to fragments, or even solitary LTRs. This may explain the origin of retrotransposons. These come in two types: long and short interspersed repetitive elements (LINEs and SINEs), and it now appears likely that they are heavily degraded fragments of ancient viruses.

As for positive selection, this can be readily confirmed by looking for viral genes or regulatory sequences that have been conserved and become an integral part of the human genome. We now know of many such sequences.

The first to be discovered is the remnant of a retrovirus that invaded the primate genome a little less than 40 million years ago and gave rise to what is known as the W family of ERVs. The human genome has roughly 650 such integrations. One of these, on chromosome 7, contains a gene called syncytin-1, which codes for a protein originally used in the virus's envelope but now critical to the functioning of the human placenta. Expression of syncytin-1 is controlled by two LTRs, one derived from the original virus and another from a different retrovirus called MaLR. Thus we have a quintessential viral genetic unit fulfilling a vitally important role in human biology.

Virus genes

There are many more examples. Another gene producing a protein vital to the construction of the placenta, syncytin-2, is also derived from a virus, and at least six other viral genes contribute to normal placental function, although their precise roles are poorly understood.

There is also tentative evidence that HERVs play a significant role in embryonic development. The developing human embryo expresses genes and control sequences from two classes of HERV in large amounts, though their functions are not known (Virology, vol 297, p 220). What is more, disrupting the action of LINE retrotransposons by administration of the drug nevirapine causes an irreversible arrest in development in mouse embryos, suggesting that LINEs are somehow critical to early development in mammals (Systems Biology in Reproductive Medicine, vol 54, p 11).

It also appears that HERVs play important roles in normal cellular physiology. Analysis of gene expression in the brain suggests that many different families of HERV participate in normal brain function. Syncytin-1 and syncytin-2, for example, are extensively expressed in the adult brain, though their functions there have yet to be explored.

Other research groups have found that 25 per cent of human regulatory sequences contain viral elements, prompting suggestions that HERVs make a major contribution to gene regulation (Trends in Genetics, vol 19, p 68). In support of that, HERV LTRs have been shown to be involved in the transcription of important proteins. For example, the beta-globin gene, which codes for one of the protein components of haemoglobin, is partly under the control of an LTR derived from a retrovirus.

The answer to our paradox is now clear: the human genome has evolved as a holobiontic union of vertebrate and virus. It is hardly surprising that researchers who have made these discoveries are now calling for a full-scale project to assess the contribution of viruses to our biology (BMC Genomics, vol 9, p 354).

It is also probable that this "virolution" is continuing today. HIV belongs to a group of retroviruses called the lentiviruses. Until recently virologists thought that lentiviruses did not endogenise, but now we know that they have entered the germ lines of rabbits and the grey mouse lemur. That suggests that HIV-1 might have the potential to enter the human germ line (Proceedings of the National Academy of Sciences, vol 104, p 6261 and vol 105, p 20362), perhaps taking our evolution in new and unexpected directions. It's a plague to us - but it could be vital to the biology our descendants.

http://www.newscientist.com/article/mg20527451.200-i-virus-why-youre-only-half-human.html?full=true