Drosophila As A Model For Comparative Genomics
Proposal for Drosophila as a Model System for Comparative Genomics
Andrew
Clark,
ac347@cornell.edu
June
9,
2003
Greg
Gibson,
ggibson@unity.ncsu.edu
Thom Kaufman, kaufman@bio.indiana.edu
Bryant McAllister, bryant-mcallister@uiowa.edu
Eugene Myers, gene@eecs.berkeley.edu
Patrick O’Grady, ogrady@amnh.org
Introduction
The challenge of obtaining a complete annotation of functional genes and regulatory elements in the genomes of
higher organisms remains a rate-limiting step to biological discovery. Recently, impressive progress has been reported based
on comparative analysis of genome sequences of related species (Boffelli et al. 2003, Kellis et al. 2003). These studies
underscore the pressing need to establish a model system for developing and applying sophisticated comparative algorithms
directed at problems of whole genome annotation. Many attributes of genomes are shared across species, so there is great
potential in applying interspecific contrasts to identify genomic features with varying degrees of conservation. For example,
the use of comparative information will accelerate the verification of gene coding regions and the discovery of regulatory
regions. To date, this has mostly been accomplished by relatively simplistic application of large-scale alignments using
MultiPIP or VISTA. It seems time to develop a model system for designing next-generation computational tools that will
result in a far more powerful application of comparative approaches. Indeed, NHGRI’s mammalian ENCODE project
requires the development of such tools. This development is best carried out in systems with smaller genomes that
nonetheless share many features with mammalian systems. The central argument of this white paper is that Drosophila
species provide an ideal target for inculcating and accelerating such advances in comparative genomics leading not only to a
complete annotation of the D. melanogaster but also the human genome.
There is little question that comparisons among different species provide insight into those regions whose sequence
conservation reflects evolutionarily conserved function. Traditional wisdom has been that divergent pairwise comparisons
(e.g., D. melanogaster vs. D. pseudoobscura) were appropriate for this task (Bergman et al. 2002). However, recent analysis
(Boffelli et al. 2003) suggests that multiple comparisons among species at various levels of divergence or “phylogenetic
shadowing" may provide better resolution by (a) maximizing evolutionary change across all regions of the genome, and (b)
minimizing alignment problems resulting from saturation in more rapidly evolving regions. Consider, for example, an
alignment of the human and mouse genomes. Those regions that are unchanged are likely to be under significant functional
constraint. Sequences that are difficult to align due to saturation (upstream regions, for example) are still critical to the
proper function of the genome and may be under constraint in a much shallower time scale. The idea of phylogenetic
shadowing is to obtain genomic sequence from multiple species at varying levels of divergence. By combining data from the
various species comparisons, one should have enough mutational hits to identify what is and is not significantly conserved.
In this case the identification and alignment of both slowly and rapidly evolving orthologs is straightforward because
divergence at multiple time points has been sampled.
We advocate a modification of phylogenetic shadowing to annotate genome data. The “ladder and constellation
approach,” where the ladder rungs are the various points of divergence and the constellations are clusters of species attaching
to these points, improves upon phylogenetic shadowing by offering multiple points of comparison (constellations) from the
same divergence point (ladder). More sophisticated models can be applied that contrast rates of substitution at synonymous
vs. nonsynonymous sites, or that contrast upstream substitutions that implicate changes in transcription factor binding, DNA
conformation, or other potentially important regulatory features. The approach advocated here ameliorates the problem of
“saturation” at synonymous sites by sampling from multiple species, including a breadth of phylogenetic diversity. Thus, the
power to obtain unambiguous counts of substitutions keeps the analysis clean, while statistical power is gained by adding
additional species (Anisimova et al. 2001, 2002).
These results make a compelling case for the complete sequencing of constellations of closely related species
selected on the basis of phylogenetic information, suitability of species, and potential use to the broader community. We
propose that by considering full genomic sequences from multiple species, it will be possible to develop annotation methods
that make full use of the latent information in patterns of interspecific divergence. The proposed goals are to:
1. Use
D. melanogaster and select taxa in the genus Drosophila as a model system to learn rules for optimizing the
selection of species and to guide the development of computational tools for full annotation of the complex
genomes.
2. Combine evolutionary information with genetic analysis in a representative series of Drosophila species to
make biological sense of whole genome sequence data.
3. Characterize the rates and patterns (and by inference, the selective forces) underlying evolutionary divergence,
which may in turn lead to speciation, across the genomes of multiple species.
4. Explore the relationships between genotype, developmental pattern and process, and the resultant phenotype of
higher eukaryotes.
5. Study the emergence and loss of new pathways, the gain and loss of complexity of pathways, and the changes in
the regulatory network of complex genomes.
Opportunities in Bioinformatics Realized by Comparative Genomics
A primary thrust of this project would be to establish a close collaboration between biologists and bioinformaticists
to develop the next generation of computational tools for automated annotation and analysis of multiple genomes in an
evolutionary context. The special issues that arise in bioinformatics in the context of comparative genomics include:
• Assembly – Are there unique features of collections of sequence data among related organisms that allow more
efficient and accurate genome assembly?
• Gene finding – Once the alignment is in hand for several species, what are the best algorithms for gene finding and
for annotating the structural features of genes?
• Regulatory sequences – How accurately can regulatory regions be annotated based on the alignments of gene
regions from related taxa?
• Transposable elements– What can a full annotation of appearance, disappearance, expansion and contraction of
transposable element families reveal about the evolution of mobile elements and consequences for genome
evolution?
• Small RNAs – What is the role of micro RNAs in gene expression? Can these be more readily detected via
comparative genomic analysis?
• Adaptive evolution – Is it feasible to make inferences about adaptive evolution across the entire genome?
• Functional validation – How might the potential value of an evolutionary change in gene or genome structure on
biological function be rigorously assessed?
In summary, the completion of genome sequences of multiple Drosophila species is motivated by: 1) the need to develop
tools for annotation by incisive comparative analysis; 2) the desire to fully annotate and validate the model organism,
Drosophila melanogaster; and 3) to stimulate development of efficient bioinformatics approaches that can be applied to
elucidate genome function and evolution beyond model systems to all higher organisms including humans.
Choice of additional Drosophila species for genome sequencing
The small genome size of most Drosophila species, the excellent phylogenetic work that has been done on this
genus, and the knowledge base of the biology of drosophilids in general allow us to select a set of species for whole genome
sequencing that are optimal with respect to the general problems in annotation cited above. Importantly, it will be possible to
perform this multiple species analysis at a fraction of the cost of sequencing a single mammalian genome. Another benefit of
using Drosophila as a comparative genomic model is that an active community, working via the Flybase interface, has
already greatly assisted the annotation of D. melanogaster and D. pseudoobscura genomes. Arrangements for similar
annotation capability will be made available to streamline the annotation of the additional genomes proposed here (see letter
of support from W. Gelbart).
Our recommendations of candidate taxa for whole genome sequencing can be divided into three broad classes based
on their relationship to D. melanogaster (Figure 1), the main goals and expected benefits of the proposed sequencing, and the
level at which sequencing will be undertaken. The first class includes taxa that are closely related to D. melanogaster and the
melanogaster subgroup. These species (D. simulans, D. yakuba, D. erecta, and D. ananassae) represent increasingly
divergent lineages that have been independent of D. melanogaster for ~5-15 million years. These taxa should be sequenced
at the 8x level.
2
Figure 1. Phylogenetic relationships and estimated
divergence times of major lineages in the genus
Drosophila. Taxa in bold are proposed for
sequencing at this time; boxed taxa have nearly
complete genome sequences. Thickened lines in D.
mojavensis and D. grimshawi indicate clades of
greater than 100 or 1,000 species, respectively.
While the Hawaiian Drosophila lineage diverged
from the virilis-repleta clade ~32 million years
ago, this lineage began to rapidly diversify 26
million years ago (hatch mark). Divergence times
are based on a linearized Adh molecular clock
(Russo et al., 1995; reviewed in Powell, 1997).
The addition of these four genomes will
also further studies of population and developmental genetics within this group through the discovery of cis-regulatory
regions and developmental motifs in rapidly evolving genes thought to be involved in speciation. The second class of species
includes taxa that are more distantly related to the melanogaster group, having been evolving independently for ~35-50
million years. This more divergent set of species (D. willistoni, D. grimshawi, D. virilis, and D. mojavensis), all of which
display behaviors and ecological attributes that are significantly different from those of D. melanogaster and each other, will
complete the series of species used in the comparative annotation of D. melanogaster. These species should also be
sequenced at the 8x level. Comparisons with these more distantly related species, which display significant rate
heterogeneity with respect to D. melanogaster, will identify constrained elements evolving at a variety of rates in the
genome. Furthermore, this entire series of species will serve to elucidate new genes, gene functions and gene expression
patterns. A better appreciation of whole genome evolution over a 60 million year time frame will result from this work,
particularly in the identification of localized synteny, which will yield a better understanding of local neighborhoods of gene
regulation. New insights will also be obtained in molecular evolutionary research through an enhanced understanding of the
evolution of codon bias and transposable elements from closely related species.
The third class, which includes two species that are closely related to D. melanogaster and D. pseudoobscura (D.
sechellia and D. persimilis, respectively), should be sequenced at the 3x level. The analysis of these sequence data will be
used to further genetic research into the mechanisms of speciation in the D. melanogaster and D. pseudoobscura model
systems, in part by establishing efficient means for developing marker loci throughout the genome and identifying candidate
genes in regions of interest. All species being considered here have a high degree of biological interest and have been used
for many years as model systems in evolutionary biology, population genetics, and physiology. Whole genome sequences
will stimulate diverse biological investigations while uniting them through well annotated and curated databases. In addition,
practical issues considered for determining the target species for sequencing included, ease of sequencing, the ability to
maintain healthy laboratory cultures, and the possibilities for experimental manipulations, including the generation of
interspecific hybrids.
Class One – Closely Related Species
We note that a proposal to sequence the genomes of D. yakuba and D. simulans was the subject of a previous White
Paper (Begun and Langley), and it is our understanding that this project has been approved, so that our proposal leverages
those data even further. The additional species recommended for sequencing are listed below with rationalizations for their
inclusion. The relative phylogenetic positions of those species chosen is shown in Figure 1 and their genome sizes and other
relevant data are given in Table 1 below.
Drosophila ananassae
Drosophila ananassae is a member of the melanogaster species group, a clade of about 300 species. The ananassae
subgroup is basal within the melanogaster group (O’Grady and Kidwell 2002) and should serve as a midpoint for
comparisons between the completed D. melanogaster and D. pseudoobscura genomes. There is a large literature available
for this species (Nishinokubi et al., 2003; Tobari et al. 1993; Yamada et al., 2002) and an active research community,
3
particularly in Japan. A characteristic of D. ananassae unique among Drosophila species is that males undergo a significant
level of recombination (Matsuda et al. 1993; Sato et al., 2000). The data available suggest that the genome of D. ananassae
is similar in content and size to that of D. melanogaster and D. pseudoobscura (Powell 1997), making whole genome
sequencing technically feasible.
Drosophila erecta
Drosophila
erecta is basal within the melanogaster subgroup and has increasingly been used as an outgroup for
comparisons between various species within this subgroup (Arhontaki et al., 2002; Parsch et al., 2001). Bergman et al.
(2003) have generated genomic sequences from 8 regions (ca. 500 kb) from D. erecta to determine their utility in aiding the
annotation of the D. melanogaster genome. They found that adding additional genomic sequences can greatly inform
annotation and can aid in the discovery of cis-regulatory regions and changes in microsynteny. Some data suggest that the
genome of D. erecta, while about the same size as that of D. melanogaster (Table 1), possesses far fewer satellite repeats,
potentially making sequencing of the entire genome of this species easier. Highly inbred lines for this species already exist.
Class Two – Divergent Taxa
The genus Drosophila contains over 2000 species, arranged into about 50 species groups. Perhaps the best studied
of these are the virilis, repleta, and Hawaiian picture wing species groups, each of which have adapted to a significantly
different ecological niche, offering the opportunity to investigate new gene pathways and functions.
Drosophila willistoni
The
willistoni species group is a Neotropical lineage within the subgenus Sophophora, a larger group containing
both the melanogaster and obscura species groups (Figure 1). In addition to being independent of the D. melanogaster
lineage for ~30 million years, the rate of evolution in the willistoni lineage is much accelerated relative the melanogaster
lineage, making this species critical for annotation of more conserved genes. Polytene chromosome polymorphism within D.
willistoni has been examined extensively, and a particularly active research community, in both the United States and Latin
America, exists. Drosophila willistoni is a significant choice for sequencing in that it has been implicated in the horizontal
transfer of the P transposable element into the D. melanogaster genome and additional components of these genomes may
have also transferred in this manner. The genome size of D. willistoni is not significantly larger than that of D. melanogaster
(Table 1), making sequencing time and effort reasonable. Moreover, a strain is now in the fourth generation of single
brother-sister mating to develop an inbred, homokaryotypic strain.
Drosophila virilis
The
virilis species group is considered as retaining many ancestral characteristics of primitive Drosophila, and
therefore, is recognized as an excellent outgroup to the Sophophoran lineage. Drosophila virilis is one of the most thoroughly
studied Drosophila species, in terms of genetic mutants (Alexander 1976), detailed linkage and physical maps (Gubenko and
Evgen’ev 1984), and gene sequences (~100 different single-copy gene sequences in GenBank). Although the total genome
size of D. virilis is about twice as large as D. melanogaster, the euchromatin appears to be only about 20% larger (Bergman
et al. 2003). Indications are that the D. virilis genome may contain more regions of simple repetitive DNA interspersed in
euchromatin than encountered with D. melanogaster. At 8x coverage, the resulting assembly of the D. virilis sequence may
be more fragmented; however, nearly 100 markers are mapped to the polytene chromosomes and these will facilitate
organization of the resulting sequence scaffolds. An existing set of mapped P1 clones could be used to further assemble the
sequences (Lozovskaya et al. 1993; Vieira et al. 1997). The community of Drosophila researchers interested in
heterochromatin has expressed strong support (see letter from S. Elgin and 22 others) in obtaining the genome sequence of D.
virilis, and the findings resulting from their efforts on this species will increase the understanding of genomes with more
complex organizations, such as the human genome. Inversions are not present in most lines of D. virilis, and many highly
inbred strains are available for sequencing.
Drosophila grimshawi
The Hawaiian Drosophila lineage represents one of the most astounding radiations of morphological and biological
complexity. This lineage consists of roughly 1,000 species, all of which are descended from a single ancestral migrant that
arrived in the Hawaiian Islands ~26 million years ago (DeSalle 1992; Russo et al. 1995). The historical fragmentation of the
Hawaiian Islands makes this group an excellent model for species formation and population genetics. Moreover, Hawaiian
drosophilids are generally large in size, often several times larger than D. melanogaster, and are characterized by elaborate
morphological features (Figure 1). Therefore, a complete genome sequence of this lineage will provide a stimulus for
4
studies of the evolution of development in addition to its usefulness in comparative annotation. The picture wing species
group has served as an evolutionary model system of speciation, sexual selection, and adaptive radiation for over 40 years,
and much is known of its behavior, ecology, and chromosome evolution. Carson (1992) used D. grimshawi as a standard to
infer phylogenetic relationships of nearly all described picture wing species. Drosophila grimshawi is therefore an excellent
candidate for obtaining a genome sequence for a representative of the Hawaiian Drosophila. An isochromosomal line, which
has been maintained in culture for nearly 40 years, could be used to sequence the D. grimshawi genome (Table 1).
Drosophila mojavensis
A fourth independent lineage representing distant comparisons with D. melanogaster is provided by D. mojavensis,
a species in the repleta group. The repleta species group contains over 100 species, all of which are native to the Americas.
In addition to being appropriately positioned for comparative bioinformatics, the repleta group provides unique biological
aspects for which a genome sequence will facilitate more detailed studies. Three species in particular, D. hydei, D. buzzatii,
and D. mojavensis, have been used as genetic and evolutionary model systems and each have a number of inbred lines and, in
some cases, libraries. Of these species, D. mojavensis emerges as the most appropriate candidate for sequencing because it is
highly specialized on its host plants and is in the process of diversification. As such, it can provide unprecedented insights
into the genetics of two primary evolutionary processes, speciation and specialization. Drosophila mojavensis, unlike D.
melanogaster and its relatives, easily hybridizes and introgresses with its close relative, D. arizonae, facilitating genetic
studies (Ruiz et al, 1990; Pantazidis et al 1993). In addition, it is, itself, in the process of speciating, with different
geographic host races exhibiting genetic differentiation and reproductive isolation (Markow et al 2002). This species also
specializes on particular host cacti endemic to the inhospitable Sonoran Desert, and offers the opportunity to address issues of
adaptation to harsh dietary chemicals (Fogleman and Danielson 2002) and to thermal extremes (Stratman and Markow 1998).
At this time a large microsatellite library is available (Ross et al 2003) that is being used to construct a genetic map for D.
mojavensis. A cDNA library has also been constructed for use in producing a microarray. Drosophila mojavensis is
sufficiently closely related to other repleta species, such as D. buzzatii, to take advantage of existing genetic and linkage data
to assist in sequence assembly and interpretation. The unique position of D. mojavensis will enable investigations into the
speciose repleta group.
Speciation Studies in the D. obscura and D. melanogaster Groups
Drosophila sechellia
Within
the
melanogaster subgroup, the sibling species triad including D. simulans, D. sechellia, and D. mauritiana
has been extensively studied in an attempt to understand the population genetic and other factors that might lead to
speciation. Drosophila simulans, a human commensal, has the largest distribution within this group. Initially, this species
was found in sub-Saharan Africa, along with some other members of the melanogaster subgroup (D. melanogaster and D.
yakuba). Drosophila sechellia and D. mauritiana are endemic to Seychelles and Mauritius, respectively. Moreover, in
addition to its restricted distribution and smaller population sizes, D. sechellia is also known to breed in the highly toxic fruit
of Morinda citrifolia (Rubiaceae). Therefore, D. sechellia is not only an excellent model for species formation, but
ecological adaptation as well, and provides an essential counterpoint to D. melanogaster for interpretation of the effect of
recent population expansion on genetic diversity.
Drosophila persimilis
Dobzhansky’s work on D. pseudoobscura and its sister species, D. persimilis, has become one of the paradigms of
modern population genetics (Anderson et al. 1991; Jones et al. 1981; Popadic and Anderson 1994). With the near completion
of the D. pseudoobscura genome, additional studies into the population genetic basis of species formation are now possible.
This work will serve as an important comparison to similar studies in the melanogaster subgroup. Like D. sechellia, we
propose a shallow shotgun sequence (3x) of the D. persimilis genome, assembled on the D. pseudoobscura backbone, to
further enable such work. There are several stocks available from the Tucson Stock Center, some of which contain
phenotypic mutants (Table 1). The establishment of a highly inbred line suitable for sequencing would be trivial for this
species and its genome size makes it quite feasible for sequencing and assembly.
Table 1. Summary of Criteria Used to Assess the Proposed Species
5
Genome
No.
Inbred
Species
Sizea
Stocksb
Geneticsc
Lined
Crossablee
D. ananassae
0.38-0.39 15
Y
Y
Y
D. erecta
0.31-0.32 1
N
Y
N
D. willistoni
0.52 20 Y Y
Y
D. grimshawi
0.47-0.53 5
Y
Y
Y
D. mojavensis
0.38-0.48 30
Y
Y
Y
D. virilis
0.62-0.68 54
Y
Y
Y
D. persimilis
0.34-0.39 14
Y
Y
Y
D. sechellia
0.34-0.35 14
Y
Y
Y
aGenome size is in picograms. bNumber of wild type and mutant stocks available from Bloomington, Tucson, or other
laboratories. cIndicates that there is a history of genetic crosses with the species and, in many cases mutant lines are
available. dThere is either an inbred line suitable for sequencing at this time or such a line will be available by August
2003. eWill form fertile hybrids with closely related species, allowing for interspecific genetic studies.
Sequencing of an outgroup
Although the strategy of phylogenetic shadowing concentrates on a group as a whole, it should be pointed out that
for many evolutionary analyses an outgroup species is essential. Within subsets of the species targeted for this proposal,
some will provide outgroups for others, e.g., D. willistoni can serve as an outgroup for the melanogaster species group and
members of the subgenus Drosophila can serve as an outgroup to the willistoni-melanogaster comparison. For the study as a
whole however there would be great value in sequencing a more remote relative. We consider this task outside the domain of
this proposal, but suggest two possibilities for future efforts. The first would be a species from another drosophilid genus of
such as Chymomyza or Scaptodrosophila. Both are well outside genus Drosophila and would be suitable outgroups for any
evolutionary comparisons within that group. Alternatively, a more distant Acalyptrate Dipteran would bridge the
evolutionary distance between the drosophilids and mosquitoes. One possible choice, in view of its taxonomic position,
economic importance and research community, would be the medfly Ceratitis capitata (family Tephritidae). Data from any
of these groups would help bridge the gap between the extant Anopheles gambiae mosquito sequence and the body of
information that would be obtained from the group of Drosophila species proposed here.
Stimulus to radical advances in bioinformatics
The use of a collection of related genomes to better annotate each individual genome is a primary goal of
comparative genomics, but a second, equally exciting goal is to learn more about how genomes evolve. The proposed project
using drosophilids provides unprecedented opportunities to accomplish these goals. In both cases, it is essential to sample
genome sequences from species that are at various levels of divergence, such that the maximum number of genomic regions
can be annotated. Such an arrangement is necessary for genome-wide comparisons, because different parts of a genome
evolve at different rates. In this proposal, this “ladder” is represented by D. melanogaster, D. simulans, D. yakuba, D.
ananassae, D. pseudoobscura, D. willistoni, and the more distantly related species in the subgenus Drosophila (D.
grimshawi, D. mojavensis, D. virilis). From such an evenly distributed set of species one can understand the rate and nature
of evolutionary events. Genome sequences in the “ladder” pattern of increasing evolutionary distance define more and more
functionally important sequences as differences accumulate over evolutionary time. For example, a rapidly evolving gene or
enhancer is more likely to be detected among the near-species, whereas detection of a housekeeping gene will likely be
facilitated by the distant-species.
In addition to having a ladder of species, having a “constellation” of species that are equally diverged provide
information in a manner similar to phylogenetic shadowing. Genome sequences in the “constellation” pattern of similar
distances in independent evolutionary lineages emphasize features by highighting the intersection of mutually conserved
regions that may be functionally important. For evolution, one can better see which elements of genomic organization are
essential as opposed to coincidental. In the proposed suite of species, there are two major constellations representing two
different levels of divergence: 1) melanogaster- yakuba-erecta and 2) melanogaster-pseudoobscura-willistoni-virilis-
6
grimshawi-mojavensis. Finally, two sibling species pairs – D. sechellia & D. simulans, and D. persimilis & D.
pseudoobscura –provide the opportunity to investigate the mechanisms of speciation, and how this process influences
divergence of gene structure and function throughout the genome.
The proposed sequencing program will provide a dataset of unparalleled utility, both in terms of the development of
comparative informatics software and biological investigations. Drosophila genomes are compact and feature rich, and, at
1/30th the size of the human and mouse genomes, they are significantly less expensive in terms of computer time for
assembly and annotation. This allows investigators to focus on the analysis, rather than merely the logistics of achieving the
computation, so that more sensitive and complex computations can be undertaken. The techniques and software that results
will be of general utility, and lessons learned and tools derived from this data set can be applied to investigation on plants,
mammals, and other comparative efforts. For example, annotation remains the most time consuming aspect of genome
biology. The development of broadly applicable tools will be a service to the research community well beyond that provided
by a deeper understanding of Drosophila.
By the very nature of large mammalian genomes it will be a number of years before we acquire greater than 15
genomes in the mammalian clade with close to complete coverage. Yet investigation into comparative whole genome
techniques at the structural (e.g., assembly), mechanistic (e.g., protein coding genes, structural RNA genes, cis-regulation)
and adaptive (e.g., positive selection on coding regions) levels should be encouraged. Moreover, without widespread sample
datasets, both population theoretical motivated approaches and sequence alignment based bioinformatics methods will lag
behind the appearance of mammalian data. Experience shows that it typically takes two to three years for new algorithms and
software to appear for any new domain. Thus in a scenario without an early developmental set of metozoan genomes it is
possible that the mammalian clade data will not deliver on the clear promise of unraveling many of the aspects of human
cellular and organismal biology in a timely fashion.
Annotation methods based on multiple aligned genome sequences can draw on evolutionary information in the form
of a broad range of conservation levels to greatly increase accuracy and utility of annotation (Cooper 2003). For a given
feature, models can be created for all genomes subject to the constraint that the predictions have the same structure or
homology. Consider for example finding protein-coding genes. The most powerful way to find genes comparatively is to
train Hidden Markov Model (HMM) predictors (e.g., Genescan) on each genome using available cDNA verified genes as the
training set, and then apply each HMM to its genome, subject to the constraint that all predicted proteins whose similarity to
each other is consistent with the known evolutionary relationship among the genomes. Note that this approach is symmetric,
since a prediction is made for all genomes, not just one target genome. Current methods are not ideal, as they are still based
on the idea of masking a reference genome with the segments that are conserved in other genomes. Loss of sequence
conservation has also been effectively used, somewhat surprisingly, within closely related species, i.e., phylogenetic
shadowing (Boffelli et al. 2003). The proposed data set would allow for the development of methods that simultaneously
exploit shadowing and footprinting.
This same idea applies to non-translated genes such as structural RNAs, where the constraint is that all predictions
simultaneously have the same or nearly the same secondary structure (Eddy 2002). While just the presence of secondary
structure in one genome is insufficient, conservation across several is likely to be informative at certain evolutionary
distances. Similarly for core promoters and enhancers, a set of position weight matrices can be created that preserve the
number and concentration of sites in the enhancer region, and allow variation in order and orientation across species. It is
further conjectured that the very close-in species will be needed for finding these types of elements.
While various nascent software tools have aligned several large genomes, it is hardly the case that this is a solved
problem. For example, a recent issue of GENOME BIOLOGY dedicated to the comparison of the human and mouse genomes
demonstrates this. All three analytical methods used (Schwartz et al. 2003, Bray et al. 2003, Couronne et al. 2003) employed
different heuristics and resulted in somewhat discordant results. Development of better methods is needed and having 6-12
complete and related genomes in a spectrum of divergence will not only improve the pair wise case but may also lead to
methods that explicitly take advantage of the known evolutionary relationship among the genomes.
Except for the sibling species pairs created by sequencing D. sechellia and D. persimilis, we argue for 8x shotgun
sequencing in order to have independently assembled sequences that are nearly complete. Whole shotgun assemblers are
now available that would reconstruct the euchromatin with perhaps 1,000 or so internal, small gaps and further provide a
good picture of the non-repetitive portions of the heterochromatin. With less than 5x shotgun sequencing, one is forced into
mapping reads to a known genome in order to assemble the data. In such an approach most of the information about
translocations, duplications, and deletions is lost. While individuals who wish to view the genome as an unordered sack of
proteins may be content with such a result, anyone interested in genomic evolution, the diaspora of retrotransposons, or cis-
regulation, to name but a few topics, will find the results inadequate. With 5-6x one can assemble independently, but one
will invariably be missing a portion of a gene such as a bit of an exon, core promoter, or enhancer region. If we truly want to
7
provide a dataset that is high value resource to both experimental and computational biologists, it seems wise to let the
sequencing pipelines run to 8x.
Biological Rationale for the Project
Functional annotation is at least as important as structural annotation. Functional annotation begins with definition
of the molecular, cellular, and organismal roles of each and every gene in a genome, but extends much more broadly to
include: understanding of the ecological and phylogenetic context within which genes function; assembling the structure of
biological pathways and networks of interaction through which the parts assemble into a whole; and investigating the
relationship between genetic and phenotypic variation both within and between species. As the 1/30th scale model for
annotation of the human or mouse genome, Drosophila provides exceptional opportunities for biological interpretation of
genomic diversity. In this section, we describe how whole genome sequencing of the selected Drosophila species will
advance biomedical research through its impact on functional annotation of the human genome, the Anopheles genome, and
of course the genome of D. melanogaster.
The rate-limiting step in functional annotation of genes has always been the availability of mutant stocks. No
vertebrate species offers a capability approaching that of D. melanogaster for rapid and efficient validation of functional
predictions in relation to individual genes, or for forward genetic generation of phenotypes for unannotated genes.
Deficiency stocks will soon cover the entire genome, and P-element mutations are expected to soon exist for more than 80%
of all fly genes (Spradling et al. 1999; Peter et al. 2002). Numerous mutant stocks also exist for several other fly species, and
modern transposon-mediated transformation methods as well as RNAi techniques enable genetic analysis for each of the
other Drosophila species (Atkinson and James 2002; Giordano et al. 2002; Brown et al. 1999). As genome comparisons
identify genes that are either absent or significantly altered in particular lineages, the molecular genetic tools of Drosophila
will facilitate direct assessment of the particular genes with both gain- and loss-of function assays. Microarray tools and
proteomic methods are in place that will allow comparative analysis of gene and protein interactions in related species, and
the wide range of cell and developmental biological tools for hypothesis-driven analysis of gene function in Drosophila need
not be described in detail here. The majority of human disease-associated genes have orthologs in the Drosophila genome
(Rubin et al. 2000), and the ability to compare function and interactions in a variety of species contexts will undoubtedly
further enhance the attractiveness of Drosophila for annotating human, not just fly, genes.
Comparative sequence analysis has been recognized as a core element of the human genome project since its
inception, and is now embraced by the commitment to sequencing of almost a dozen chordate genomes in the next few years.
These species will cover the range of phylogenetic organization from primate through mammal, and tetrapod to primitive
vertebrates. Useful as these sequence data will be, there is no question that comprehensive functional dissection will be
required to make sense of it. Furthermore, there are more complex questions that cannot be addressed purely with
bioinformatic tools, including: 1) how can function be ascribed to the one third of the predicted proteome that simply returns
“function unknown” query results; 2) what methods will enable characterization of functional divergence of the conserved
fraction of genes; 3) to what extent are developmental and physiological pathways conserved across species, and what
empirical approaches can be used to detect such conservation; 4) what are the effects of structural factors such as
chromosomal location, heterochromatin, transposable element movement, and nucleotide content bias on gene function and
evolution; and 5) how has recent speciation followed by rapid population expansion and migration shaped the genetic risk
factors that contribute to complex diseases such as diabetes, heart disease, psychological disorders and neurological decline
with age. As described above, the genus Drosophila provides an obvious model for addressing each of these questions, and
the various species have been selected with these questions in mind.
A considerable investment has been made in sequencing of the genomes of malaria-transmitting mosquitoes. Like
other insects that directly impact public health as carriers of disease, the best hope for functionally annotating and potentially
manipulating the genomes of these species lies in detailed understanding of Drosophila genetics (Zdobnov et al. 2002).
Whole genome sequencing will facilitate this process, for example by allowing mutational analysis of homologs and
transgenic experimentation. It is also sure to promote renewed interest in Drosophila as a model for host-parasite
interactions. The interaction between the bacterium Wolbachia and D. melanogaster is well known (e.g., Rand et al. 2001),
but various Drosophila species including those included in this proposal are also host to nematodes and/or form symbiotic
relationships with a range of microbes.
The biological basis of species differences
In addition to providing detailed annotation for the D. melanogaster genome, this project would identify genes and
genomic features that are unusually highly diverged, potentially because of natural selection for diverse function. Critical to
this analysis are sequences with sufficient divergence to detect a signal, yet not so divergent that multiple mutations have
8
obscured the actual signal. Both likelihood and Bayesian methods have been devised to estimate parameters of explicit
evolutionary models for both coding and noncoding regions in a series of increasingly divergent species. These methods are
able to identify the genes that exhibit strongly selected patterns, implying that these genes exhibit evidence for directional
selection and hence functional differentiation between species (e.g., Yang and Nielsen 2000, 2002, Yang et al. 2000, see
Swanson et al. 2001 for a specific example). The D. melanogaster - D. pseudoobscura species pair is already of an
evolutionary distance where synonymous sites in coding regions are saturated for mutations (e.g., Table 2 of Bergman et al.
2003). However, with a series of additional intermediate sequences, the stage is set for the genome-wide application of
likelihood analysis of genome features that carry adaptive attributes. In short, this will be a remarkable resource for
understanding the nature of adaptive differences between closely related species.
Impact on the Drosophila research community
Worldwide, there are well over one thousand active Drosophila research groups that are in a position to capitalize
immediately on the data obtained from the proposed project. This results in a total volume of research annually that is similar
to the mouse community, and dwarfs the magnitude of research conducted on any other model organism. Many of the groups
already use comparative data to guide their annotation of gene structure, and it is noteworthy that the Tucson Species Stock
Center has seen a four-fold increase in orders for D. pseudoobscura since the draft genome of this species was released two
months ago. While the bulk of fly research is focused on D. melanogaster, > 200 labs actively employ other species in their
research on the evolution of development, the biology of diverse species, and population/quantitative genetics. Indeed,
evolutionary and developmental biologists are making a substantial contribution to genome research using Drosophila
(Reinke and White 2002). Moreover, a computational research community is already in place to ensure the rapid
development of new annotation approaches and interpretative frameworks (via Flybase, etc.). As the human genome project
increasingly focuses on the genetics of complex diseases, there is every expectation that Drosophila will continue to drive
theoretical developments and understanding of the forces that shape genetic and phenotypic variation. The increasing number
of medical school-based labs that are working on Drosophila species further attests to the medical relevance of this group of
organisms. Survey of symposium and workshop titles over the past five years also testifies to dramatic expansion of
research interest from a focus on development to all aspects of cell biology, neurobiology and behavior, physiology and
biochemistry, and chromatin structure, such that Drosophila is now truly a broad biomedical research organism. Further
interest in the genetics of adaptation and divergence relating to all of these aspects of organismal biology will be guaranteed
by the generation of sequence data from species with a range of ecological and phylogenetic divergence.
Scheduling for the proposed work
Because this one proposal entails multiple species sequences, it is useful to consider the order in which the proposed
work is done. This ordering is based on the expectation of the immediate gain to be accrued from each successive genome
sequence. The next Drosophila genome to be done should be D. willistoni, because this sequence would produce the largest
initial impact in resolving ambiguities in D. melanogaster annotations even with the current software tools. The second set of
genomes to do would be D. erecta and D. ananassae, as these would provide a series needed to establish a powerful,
genome-wide phylogenetic shadowing approach, and would stimulate rapid development of bioinformatics tools for
comparative genomic annotation. Third would be D. virilis, to reach further out from all other Drosophila, serving to
annotate the most slowly evolving developmental features. Fourth would be to perform the lighter shotgun coverage of
sechellia and persimilis, because by this time the informatics tools for annotation would be ready to capitalize on the two
additional closely related species, and the evolutionary genetic community would be primed to use genome-wide
comparisons of close sibling species to investigate the genetic basis for differences between them. Finally would come D.
grimshawi and D. mojavensis, not because they are the least important, but rather because by this time these research
communities would be fully prepared to take optimal advantage of this new resource. Note that the small size of the
Drosophila genomes means that the entire span of time before acquiring all eight genomes should be between 12 and 20
months.
Conclusion
The core thesis of this proposal is that the genus Drosophila provides a superb model system for comparative
genomics. Dollar for dollar, sequencing of ten Drosophila species will provide more information than sequencing of a single
large vertebrate, because it will facilitate the development of computational tools for using comparative genomics in both
structural and functional annotation. Furthermore, the Drosophila research community is large and the tools of Drosophila
genetics are well tuned, so that any inferences of function derived from the novel bioinformatics tools described here will
rapidly face the harsh test of direct biological validation. The species to be studied parallel the diversity seen at three levels
of importance to understanding the genetic basis of human variation, namely primate, mammal, and vertebrate. The sequence
9
resources will greatly enhance fundamental experimental research into the mechanisms of speciation, the genetics of
adaptation, genotype-phenotype mapping, chromatin organization, and of course developmental and biochemical
mechanisms. Thus, we can envisage these sequences being the substrate that leads to the emergence of a myriad of
computational and empirical strategies for making sense of sequence diversity. In addition to theoretical and bioinformatics
advances, the interplay between the comparative genomic analysis of these sequences and evolutionary inferences to be
drawn from them are likely to generate a wealth of proposed hypotheses in the fields of embryogenesis, development,
neuronal molecular biology, innate immunity and life span. These will face immediate test through the excellent molecular
biology tools available in D. melanogaster. This approach has particular relevance to human genetic disorders, and there is a
growing effort to use Drosophila as a direct model for human genetic diseases (Rubin et al. 2000). This project is of
importance to the public health mission of the NIH both through the development of a powerful infrastructure of methods for
comparative genomics, and also by further empowering the Drosophila model system to make primary discoveries about
gene function and gene regulation relevant to human disease.
References
Adams MD, Celniker SE, Holt RA, Evans CA, Gocayne JD, Amanatides PG, Scherer SE, et al. 2000. The
genome sequence of Drosophila melanogaster. Science 287:2185-2195.
Alexander ML. 1976. The genetics of Drosophila virilis. Pp, 1365-1427, In: The Genetics and Biology of
Drosophila. Vol. 1c, Ashburner and Novitski (eds.), Academic Press.
Anderson WW, Arnold J, Baldwin DG, Beckenbach AT, Brown CJ, Bryant SH, Coyne JA, et al., 1991. Four
decades of inversion polymorphism in Drosophila pseudoobscura. Proc. Natl. Acad. Sci. 88: 10367-10371.
Anisimova M, Bielawski JP, Yang Z 2001. Accuracy and power of the likelihood ratio test in detecting adaptive
molecular evolution. Mol Biol Evol 18:1585-1592
Anisimova M, Bielawski JP, Yang Z. 2002. Accuracy and power of Bayes prediction of amino acid sites under
positive selection. Mol Biol Evol. 19:950-958.
Arhontaki K, Eliopoulos E, Goulielmos G, Kastanis P, Tsakas S, Loukas M, Ayala F. 2002. Functional
constraints of the Cu,Zn superoxide dismutase in species of the Drosophila melanogaster subgroup and
phylogenetic analysis. J Mol Evol. 55:745-756.
Atkinson PW, James AA 2002. Germline transformants spreading out to many insect species. Adv. Genet. 47:
49-86.
Barker, JSF, and Starmer, WT. 1982. Ecological Genetics and Evolution: The Cactus-Yeast-Drosophila Model
System. Academic Press, NY.
Barker, JSF, Starmer, WT, and MacIntyre, RJ. 1990. Ecological and Evolutionary Genetics of Drosophila.
Plenum Press, NY.
Bergman CM, Pfeiffer BD, Rincon-Limas DE, Hoskins RA, Gnirke A, Mungall CJ, Wang AM, Kronmiller B,
Pacleb J, Park S, Stapleton M, Wan K, George RA, De Jong PJ, Botas J, Rubin GM, Celniker SE. 2003.
Assessing the impact of comparative genomic sequence data on the functional annotation of the Drosophila
genome. Genome Biol. 3:Research0086-6.
10
Boffelli D, McAuliffe J, Ovcharenko D, Lewis KD, Ovcharenko I, Pachter L, Rubin EM. 2003. Phylogenetic
shadowing of primate sequences to find functional regions of the human genome. Science 299:1391-1394.
Bray N, Dubchak I, Pachter L 2003. AVID: A global alignment program. Genome Res 13:97-102.
Brown SJ, Mahaffey J, Lorenzen M, Denell RE, Mahaffey JW 1999. Using RNAi to investigate orthologous
homeotic gene function during development of distantly related insects. Evol Dev 1: 11-15.
Carson, H. L. 1992. Inversions in Hawaiian Drosophila. Pp. 407-440 In: Krimbas, CB and Powell, JR (eds)
Drosophila Inversion Polymorphism. CRC Press, Boca Raton, FL.
Cooper GM, Brudno M, Green ED, Batzoglou S., Sidow A. 2003. Quantitative estimates of sequence divergence
for comparative analyses of mammalian genomes. Genome Res. 13: 818-820.
Couronne O, Poliakov A, Bray N, Ishkhanov T, Ryaboy D, Rubin E, Pachter L, Dubchak I. 2003. Strategies and
tools for whole-genome alignments. Genome Res 13:73-80.
DeSalle, R. 1992. The origin and possible time of divergence of the Hawaiian Drosophilidae: evidence from
DNA sequences. Mol. Biol. Evol. 9: 905-916.
Eddy SR. 2002. Computational genomics of noncoding RNA genes. Cell 109: 137-140.
Fogleman, JC. 1982. The role of volitiles in the ecology of cactophilic Drosophila. Pp191-208. In: Barker, JSF,
and Starmer, WT. Ecological Genetics and Evolution: The Cactus-Yeast-Drosophila Model System. Academic
Press, NY.
Fogelman, J.C. and Danielson, P.B. 2002. Chemical interactions in the cactus-microorganism-Drosophila model
system of the Sonoran Desert. Am. Zool. 41:877-889.
Giordano E, Rendina R, Peluso I, Furia M 2002 RNAi triggered by symmetrically transcribed transgenes in
Drosophila melanogaster. Genetics 160: 637-648.
Gubenko, I.S., Evgen'ev, M.B. 1984 Cytological and linkage maps of Drosophila virilis chromosomes. Genetica
65:127-139.
Jones JS, Bryant SH, Lewontin RC, Moore JA, Prout T 1981. Gene flow and the geographical distribution of a
molecular polymorphism in Drosophila pseudoobscura. Genetics 98:157-178.
Kellis M, Patterson N, Endrizzi M, Birren B, Lander ES. 2003. Sequencing and comparison of yeast species to
identify genes and regulatory elements. Nature 423:241-254.
Markow, T.A., Castrezana, S., Pfeiler, E. 2002. Flies across the water: Genetic differentiation and reproductive
isolation in allopatric desert Drosophila. Evolution 56:546-552.
Matsuda, M., Sato, H., Tobari, Y.N. 1993. Crossing over in males. In "Drosophila ananassae. Genetical and
biological aspects." Ed. Tobari,Y.N. Japan Scientific Societies Press.
Nishinokubi I, Shimoda M, Kako K, Sakai T, Fukamizu A, Ishida N. 2003. Highly conserved Drosophila
ananassae timeless gene functions as a clock component in Drosophila melanogaster. Gene 307:183-190.
O'Grady, PM, Kidwell, MG 2002. Phylogeny of the subgenus Sophophora (Diptera:Drosophilidae) based on
combined analysis of nuclear and mitochondrial sequences. Mol. Phylogenet. Evol. 22: 442-453.
11
Pantazidis, A.C., Galanopoulos, V.K., Zouros, E. 1993. An autosomal factor from Drosophila arizonae restores
normal spermatogenesis in Drosophila mojavensis males carrying the Drosophila arizonae Y chromosome.
Genetics 134: 309-318.
Parsch J, Meiklejohn CD, Hauschteck-Jungen E, Hunziker P, Hartl DL. 2001. Molecular evolution of the ocnus
and janus genes in the Drosophila melanogaster species subgroup. Mol Biol Evol. 18:801-811.
Peter, A., Schottler, P., Werner, M., Beinert, N., Dowe, G., Burkert, P., Mourkioti, F., et al., 2002. Mapping and
identification of essential gene functions on the X chromosome of Drosophila. EMBO Rep 3: 34-38.
Popadic, A., Anderson, W.W. 1994. The history of a genetic system. Proc. Natl. Acad. Sci. USA 91: 6819-6823.
Powell, J. R. 1997. Progress and Prospects in Evolutionary Biology: The Drosophila Model. Oxford University
Press, New York.
Rand DM, Clark AG, Kann LM. 2001. Sexually antagonistic cytonuclear fitness interactions in Drosophila
melanogaster. Genetics 159:173-187.
Reinke V and White KP 2002. Developmental genomics approaches in model organisms. Annu. Rev. Genomics
Hum. Genet. 3: 153-178.
Ross, C.L., Dyer, K.A., Eerez, T. Miller, S.J. Jaenike, J., and Markow, T.A. 2003. Rapid divergence of
microsatellite abundance among species of Drosophila. Mol. Biol. Evol, May 30 Epub, PMID: 12777536
Rubin, G.M., Yandell, M.D., Wortman, J.R., Gabor Miklos, G.L., Nelson, C.R., Hariharan, I.K., Fortini, et al.,
2000 Comparative genomics of the eukaryotes. Science 287: 2205-2215.
Ruiz, A., Wasserman, M.W. and W.B. Heed. 1990. Evolution in the mojavensis cluster of cactophilic
Drosophila with descriptions of two new species. Jour. Hered. 81:30-42.
Russo, CAM, Takezaki N, and Nei M 1995. Molecular phylogeny and divergence times of drosophilid species.
Mol. Biol. Evol. 12(3):391-404.
Sato H, Goni B, Matsuda M, Tobari YN. 2000. A site specific increase in recombination in Drosophila
ananassae. Genes Genet Syst. 75:41-47.
Schwartz S, Kent WJ, Smit A, Zhang Z, Baertsch R, Hardison RC, Haussler D, Miller W 2003. Human-mouse
alignments with BLASTZ. Genome Res. 13:103-107.
Spradling A, Stern D, Beaton A, Rhem E, Laverty T, Mozden N, Misra S, and Rubin GM 1999. The Berkeley
Drosophila Genome Project gene disruption project: Single P-element insertions mutating 25% of vital
Drosophila genes. Genetics 153: 135-177.
Stratman, R. and Markow, T.A. 1998. Heat tolerance of Sonoran Desert Drosophila. Functional Ecology 8:965-
970.
Swanson WJ, Clark AG, Waldrip-Dail HM, Wolfner MF, Aquadro CF. 2001. Evolutionary EST analysis
identifies rapidly evolving male reproductive proteins in Drosophila. Proc Natl Acad Sci U S A. 98:7375-7379.
Tobari YN 1993 Drosophila ananassae. Genetical and biological aspects. Japan Scientific Societies Press
12
Yang Z, Nielsen R. 2000. Estimating synonymous and nonsynonymous substitution rates under realistic
evolutionary models. Mol Biol Evol 17:32-43.
Yamada H, Matsuda M, Oguma Y. 2002. Genetics of sexual isolation based on courtship song between two
sympatric species: Drosophila ananassae and D. pallidosa. Genetica 116:225-237.
Yang Z, Nielsen R. 2000. Estimating synonymous and nonsynonymous substitution rates under realistic
evolutionary models. Mol Biol Evol 17:32-43.
Yang Z, Nielsen R. 2002. Codon-substitution models for detecting molecular adaptation at individual sites along
specific lineages. Mol Biol Evol 19:908-917.
Yang, Z., Nielsen, R., Goldman, N., Pedersen, A.M. 2000. Codon-substitution models for heterogeneous selection
pressure at amino acid sites. Genetics 155:431-449.
Zdobnov EM, von Mering C, Letunic I, Torrents D, Suyama M, Copley RR, Christophides GK, Thomasova D,
Holt RA, Subramanian GM, Mueller HM, Dimopoulos G, Law JH, Wells MA, Birney E, Charlab R, Halpern AL,
Kokoza E, Kraft CL, Lai Z, Lewis S, Louis C, Barillas-Mury C, Nusskern D, Rubin GM, Salzberg SL, Sutton
GG, Topalis P, Wides R, Wincker P, Yandell M, Collins FH, Ribeiro J, Gelbart WM, Kafatos FC, Bork P 2002.
Comparative genome and proteome analysis of Anopheles gambiae and Drosophila melanogaster. Science
298:149-159.
13
Document Outline
