Association of microsatellite pairs with segmental duplications in insect genomes

Background Segmental duplications (SDs), also known as low-copy repeats, are DNA sequences of length greater than 1 kb which are duplicated with a high degree of sequence identity (greater than 90%) causing instability in genomes. SDs are generally found in the genome as mosaic forms of duplicated sequences which are generated by a two-step process: first, multiple duplicated sequences are aggregated at specific genomic regions, and then, these primary duplications undergo multiple secondary duplications. However, the mechanism of how duplicated sequences are aggregated in the first place is not well understood. Results By analyzing the distribution of microsatellite sequences among twenty insect species in a genome-wide manner it was found that pairs of microsatellites along with the intervening sequences were duplicated multiple times in each genome. They were found as low copy repeats or segmental duplications when the duplicated loci were greater than 1 kb in length and had greater than 90% sequence similarity. By performing a sliding-window genomic analysis for number of paired microsatellites and number of segmental duplications, it was observed that regions rich in repetitive paired microsatellites tend to get richer in segmental duplication suggesting a “rich-gets-richer” mode of aggregation of the duplicated loci in specific regions of the genome. Results further show that the relationship between number of paired microsatellites and segmental duplications among the species is independent of the known phylogeny suggesting that association of microsatellites with segmental duplications may be a species-specific evolutionary process. It was also observed that the repetitive microsatellite pairs are associated with gene duplications but those sequences are rarely retained in the orthologous genes between species. Although some of the duplicated sequences with microsatellites as termini were found within transposable elements (TEs) of Drosophila, most of the duplications are found in the TE-free and gene-free regions of the genome. Conclusion The study clearly suggests that microsatellites are instrumental in extensive sequence duplications that may contribute to species-specific evolution of genome plasticity in insects.


Background
Microsatellites are tandem repeats of simple sequences (usually consisting of motifs less than 6 bp long) which are found ubiquitously across eukaryotic genomes. Generally, microsatellite loci are assumed to be selectively neutral [1][2][3]. However, increasing evidence now suggests that microsatellites are associated with important roles in genome structure and evolution and are often subjected to selective pressure [4][5][6][7][8][9][10].
Moreover, non-random genomic distributions of microsatellites are well documented in eukaryotes [4,[11][12][13][14][15]. As much as 25% of the microsatellites are localized close to each other, generally within 10 bp, in different eukaryotic genomes as found by Kofler et al. [14]. Furthermore, Kofler et al. [14] showed that these microsatellites are localized close to each other in the genomes at a higher frequency than expected under the assumption of random genomic distribution. In addition, simple sequence coding sequences are distributed differentially in the genomes as evident from analysis of 25 insect species [10]. Also, the Mouse Genome Sequencing Consortium, [16] has revealed that the ends of chromosome arms in mice are associated with higher density of microsatellites than other chromosomal regions. However, the functional and evolutionary relevance of non-random genomic distribution of microsatellites is poorly understood [6].
Studies have indicated possible association of microsatellites with segmentally duplicated sequences in some organisms [17][18][19][20][21][22]. Segmental duplications (SDs), also known as low-copy repeats, are generally defined as DNA sequences of length greater than 1 kb which are duplicated with high degree of sequence identity (>90%) [23]. SDs are important features of genomes as they may have functional consequences in genomic instability and diseases as evident in humans [17]. SDs are generally found in the genome as mosaic forms of duplicated sequences [19]. A two-step process generates such mosaic structures [24]. In the first step, multiple duplicated sequences are aggregated at specific genomic regions. In the second step, these primary duplications undergo multiple secondary duplications. However, the mechanism of how duplicated sequences are aggregated in the first place is not clear.
The present study is a systematic investigation to determine the distribution of microsatellite sequences in segmental duplications of different insect genomes (n=20). Although microsatellites are extensively used as genetic markers for population and ecological investigations of insects [9], the relationship of microsatellite sequences with segmental duplications has not been established in spite of availability of several insect genome sequences. Here we show that specific microsatellite pairs along with the intervening sequences are repeated with different frequencies in the genome and many of the low copy repeats of these loci are segmentally duplicated, henceforth called as microsatellite-associated SDs or mSDs. The results further show that these repeated microsatellite pairs (rMP) tend to aggregate at different genome regions along with the segmentally duplicated sequences suggesting a role for microsatellites in segmental duplications in insect genomes.

Non-random association of microsatellite pairs
The SciRoKo software [25] was used to identify the mono-, di-, tri-, tetra-and hexa-nucleotide simple sequence repeats (SSRs) or microsatellites in each genome. Both perfect and imperfect SSRs were detected by using the default parameters, with fixed penalty = 5 for mismatches between motif sequences. From the output files of SciRoKo (that generates microsatellite sequences, their position in the genome with start and end coordinates in chromosomes/scaffolds/supercontigs), distances between neighboring microsatellites were calculated in each species. When two microsatellites of the same repeat motifs had the same intervening distance at more than one location, they were counted as repetitive microsatellite pair (rMP). We assumed that presence of microsatellite pairs with the same motifs and same intervening distance at multiple locations in a genome was due to a random chance. Test of this null assumption was performed by calculating statistical significance of the hypergeometric probability as follows. First, the number of microsatellite pairs associated with the same intervening distance but different SSRs (n1) and the number of pairs associated with the same SSR pairs but with different intervening distances (n2) was determined in each genome. The total number of possible combinations for these two groups of SSR pairs was calculated as C (n, n1)* C(n, n2), where 'n' is the total number of microsatellites identified in the genome minus one (i.e. interventions between microsatellites), and 'C' represents the function of combination. Thus, C(n, n1) was calculated as the number of possibilities for choosing the 'n1' pairs from all the detected microsatellites in a genome. Of these, C(n1,n3) was calculated as the number of possibilities for choosing the same SSR pairs with the same intervening distance (n3). Thus, the number of combinations of the same SSR pairs having different intervening distance was calculated as C(n − n1, n2 − n3). From these, the cumulative probability of hypergeometric distribution of SSR pairs with the same intervening distance was calcu- Thus, 1-p value provided the statistical significance to reject or not to reject the null assumption. The multiple testing by Bonferroni correction method was conducted to adjust the individual p values. The threshold values less than 0.05 were considered statistically significant unless stated otherwise. The association was further tested in shuffled sequences of A. aegypti supercontigs. Here we assumed that the distribution of SSR pairs was independent of the sequence structure of the genomic sequences and hence sequence shuffling would not affect their distribution. To test this assumption, the supercontig sequences were shuffled and sampled (n = 1,000 sequences, each of 1 kb in length) using the R code 'ShuffleAndExtract' (http://tata-box-blog.blogspot.com/search/label/R). The sequences generated from three independent shuffling experiments were then analyzed separately for distribution of rMPs using hypergeometric tests as described above.
A canonical correlation test [26] was performed using the number of rMPs associated with different intervening distances (< 10 bp, ≥ 10 bp but < 100 bp, ≥ 100 bp but <1 kb, ≥ 1 kb but < 5 kb, ≥ 5 kb but < 10 kb and ≥ 10 kb but < 50 kb) among the 20 species. Euclidean distance measures were used in the correlation test and significance of correlation was determined by permutation test (n = 9,999 random) according to methods of Anderson and Willis [26].

Intervening sequences of paired microsatellites
The intervening DNA sequences of the paired microsatellites were extracted using the coordinates of the microsatellite ends in the genome sequences by the R package SeqINR [27] or the GALAXY server (https://main.g2.bx.psu.edu/). The pair-wise alignments of duplicated sequences and the percent sequence identity of the alignments were performed using the R package 'Biostrings'. The phylogenetic analyses were conducted using the Neighbor-Joining method in MEGA4 [28]. The evolutionary distances were computed using the maximum composite likelihood method [29] and were in the units of the number of base substitutions per site. The estimates of average evolutionary divergences between different groups of rMP loci (e.g. genic versus nongenic) were also calculated by MEGA4. All the sequence polymorphism analyses including calculation of total number of mutations, number of polymorphic sites, the average number of nucleotide differences among duplicated sequences, and significance of Tajima D statistics were conducted by DnaSP v 5.10 [30].
To determine the genomic distribution patterns of paired microsatellites, genome assemblies (where sequences have been assigned to chromosomes) were binned to determine the total number of rMPs and the total number of pairs as mSDs. The size and number of bins were variable depending upon the chromosome length but they were mostly in megabases (Mb). For example, the A. gambiae genome was binned as < 1 Mb, 1-5 Mb, 5-10 Mb, 10-20 Mb, 20-30 Mb, 30-40 Mb, 40-50 Mb and > 50 Mb for each chromosome. The total numbers of rMPs and the mSDs across individual windows were counted. The Spearman rank order correlation test was performed with the total number of rMPs and the total number of mSDs among the binned regions to determine if regions rich in paired microsatellites accumulated more segmental duplications than regions poor in paired microsatellites. The p-value < 0.05 was considered significant.

Association of mSD sequences with gene duplications and transposons
The genomic positions of mSDs were used to determine if they were localized in genic regions. The start and end coordinates of annotated genes (both coding and non-coding) of each genome (Biomart dataset: Ensembl Metazoa 16) were used to determine if mSDs were localized within or overlapping with the genes. The gene ontology (GO) terms The species names are listed as four-letter abbreviations.
(downloaded from Biomart) of the genes associated with mSDs were analyzed. The rank orders of GO terms were used to determine the top ranking functions of these genes. The orthology and paralogy relationships of insect genes were obtained from Biomart (Metazoa) database. Based on sequence identity between paralogous copies, the nearly identical paralogs [31] were identified. The transposable element (TE) sequences annotated from D. melanogaster were analyzed to determine association of mSDs with transposable elements. The TE sequences were downloaded from ftp://ftp.flybase.net/genomes/aaa/transposable_elements/ ReAS/v1/consensus_fasta/. The start and end coordinates of TEs in relation to mSDs were analyzed to determine TE-mSD associations.

Identification of repetitive paired microsatellites
The microsatellite sequences were identified in a genomewide manner among the 20 insect species that included three mosquitoes, twelve Drosophila, honey bee, silkworm, beetle, wasp and pea aphid. The total number of microsatellites and the genomic density (number of microsatellites per Mb of the genome) are shown in Table 1 for each species. Collectively, the microsatellite sequences represent 0.2-2.8% of the assembled genome size of these insects. The frequencies of mono-, di-, tri-, tetra-, penta-and hexa-nucleotide SSRs in each of the 20 genomes are provided in Additional file 1. The entire list of microsatellites and their positions in the genome assembly of all the 20 species are available upon request. Based on the position of microsatellites in the genome, it was observed that specific microsatellite pairs are localized together with the same intervening distance at multiple locations in each genome ( Figure 1). We identified these repetitive microsatellite pairs (rMPs) in each insect genome comprehensively. The list of all rMPs in the D. melanogaster genome is provided in Additional file 2, but the lists of rMPs of all the 20 genomes are available upon request. It was found that in each genome the frequency of rMPs varies depending upon the intervening distance of the paired microsatellites (Table 2). Table 2 shows that rMPs of 10 bp to 1 kb in size are predominant in all the genomes. It also shows that the rMPs with length greater than 5 kb are present only in specific species. Although some rMPs are up to 50 kb in length, none of the rMPs in any genome has an intervening distance greater than 50 kb. On the other hand, rMPs of smaller sizes (< 10 bp) were relatively high in frequency in these insects (Table 2).

Non-random distribution of microsatellite pairs
We assumed that microsatellites with the same intervening distance are localized as pairs at different locations in the genome by random chance. Based on the number of paired microsatellites with same and/or different sequences and intervening distance, the distribution of the rMP represents a hypergeometric distribution as illustrated in Figure 2. We performed hypergeometric tests with Bonferroni correction as described in Methods to determine if microsatellite pairing is a random or nonrandom event. The results revealed that many of the rMPs are distributed in non-random manner (p < 0.05, Additional file 3) in the insect genomes. They account for 35 -67% of the total number of rMPs identified in the genomes. The motif sequences of the most repetitive microsatellite pairs for each species are shown in Figure 3. To further validate the result, we shuffled the genomic sequences of A. aegypti (see Methods) and repeated the hypergeometric tests on rMP distribution in the shuffled sequences. We didn't find a significant correlation in the distribution of rMPs in the shuffled sequences (data not shown). It was found that~96% of the rMPs identified from the shuffled sequences showed random distribution pattern and lacked statistical significance after the Bonferroni correction. This suggested that the genomic distribution of microsatellite pairs is not due to a random chance but may be a characteristic feature of the genome structure of the organism.

Inter-and intra-chromosomal distribution of rMPs
The inter-and intra-chromosomal repetition of microsatellite pairs were investigated in selected species where chromosome assignments of genome sequences are available (D. melanogaster, A. gambiae, N. vitripennis and A. mellifera). It was found that the inter-chromosomal duplications are generally more frequent than intrachromosomal duplications (Figure 4). However, in chromosome X and 2L of D. melanogaster, rMPs are duplicated more often within chromosomes than between chromosomes. In D. melanogaster and A. gambiae, many chromosome specific rMPs were identified (Additional file 4). These duplications are restricted to either X chromosome or autosomes but never shared by both sex chromosome and an autosome. The A. gambiae genome contains~3 fold greater number of such sequences than that of the D. melanogaster genome. However, the frequencies of these chromosome specific rMPs are relatively lower (2-10 times) compared to frequencies of rMPs shared by both sex and autosomes. Furthermore, it was found that the shared rMPs may be selected differentially between the sex chromosome and autosomes. For example, the rMP (TG)n-1022 bp-(CA)n is repeated 40 times and shared between autosomes and the sex chromosome in A. gambiae ( Figure 5). Sequence analysis showed contrasting patterns of evolution of these rMPs between the sex chromosome and autosomes indicating differential selection of sex-chromosome duplications compared to the autosomal duplications (Table 3).

rMP dependent aggregation of segmental duplications
The rMPs with intervening distances greater than 1 kb are low copy repeats in contrast to rMPs of lengths less than 1 kb (Additional file 5). We identified rMPs with intervening distance greater than 1 kb wherein sequence identity of the intervening DNA is greater than 90% [23] to identify mSDs in each species (Table 4). The Neighborjoining trees also confirmed the sequence relatedness of  The microsatellite pair M1-M2 is localized at two positions where the intervening distance is the same (d1) at both locations. The M1-M2 pair is also located at another position but with a different intervening distance (d7). On the other hand, a different pair of microsatellites (M4 and M5) is found wherein the intervening distance is again d1. M3 and M6 are not associated with any repetitive microsatellite pair. Below that, the rectangular box shows all the neighboring pairs of microsatellites identified in the genome which is basically the total counts of microsatellites minus 1. The two circles within the box contain n1 and n2 microsatellite pairs respectively, where n1 is the total count of different SSR pairs but has the same distance d1 (i.e. M4-M5 pairs) and n2 is the total counts of the same SSR pairs (M1-M2) but with different intervening distance (d7). The overlap between the two circles represents the total number (n3) of SSR pairs with the same SSR sequences (M1-M2) as well as the same intervening distance (d1). The probability of obtaining these n3 numbers of repetitive pairs of M1-M2 is the hypergeometric distribution probability shown below the circles.  these duplicated loci ( Figure 6). The sequences of duplicated copies of specific mSDs in select species are shown in Additional file 6. We performed a sliding-window analysis of the A. gambiae genome to determine the frequency of rMPs and mSDs in each chromosome. The individual chromosomes were searched in different intervals to calculate the total number of mSDs as well as the number of repetitive paired microsatellites. It was found that the frequency of mSDs increases linearly (r 2 > 78%) with the frequency of paired microsatellites in each chromosome (Figure 7). Similar results were also observed for chromosomes of other insects (Additional file 7). These results clearly suggest that regions that are already rich in rMPs tend to get richer in segmental duplications. This 'rich-gets-richer' based co-aggregation of mSDs along with rMPs suggest an rMP dependent mechanism of aggregation of duplicated sequences in the genome, as part of the two-step process of genesis of segmental duplications [24]. We then wanted to know whether higher segmental duplications with higher frequency of paired microsatellites is a withingenome feature or are they also correlated across genomes. By comparing the number of mSDs and the number of rMPs across the 20 species, it was found that the total number of segmental duplications increases significantly (Spearman correlation = 0.78, two tailed p-value = 0.00004) with increases in the total number of rMPs among the species (Figure 8) suggesting that the 'rich-gets-richer' mechanism may be a universal mode of rMP dependent aggregation of mSDs across genomes. However, it seems unlikely that the correlation between paired microsatellites and segmental duplications evolve in a phylogeny dependent manner among species. It is evident from Figure 8 that there is no correspondence between species phylogeny and the correlated variation between mSDs and rMPs.

Genic versus non-genic association of mSDs
By mapping the mSD sequences to the annotated gene locations, we identified duplicated copies which are localized within or overlapping the coding and non-coding genes (Additional file 8). These genes represented different gene ontologies (GO) in different insects (Additional file 9), among which the 'protein binding' or 'nucleus' gene ontologies represented the top ranking predicted functions across species. The genic duplications accounted for only~5-15% of the mSDs identified in different species indicating that the majority of duplications occur in the intergenic regions. The lower abundance of mSDs in genic regions compared to the non-genic regions may be related to differential selection pressure between genic and non-genic regions. For example, the duplications of (TG)n-1022 bp-(CA)n in the A. gambiae genome have genic copies (2L:47741478-47742671, 2R:4134771-4135970, 3L:10536049-10537278, 3R:12823084-12824255, 3R:13862599-13863800 and 3R:33983540-33984801) (see Figure 5). We determined the evolutionary divergence of the genic versus non-genic duplications (see Methods). It was found that the genic copies have lower average evolutionary divergence than that of the non-genic copies (0.32 versus 0.58, respectively) indicating a possibility of selection constraint on genic duplications.
In addition, we also investigated whether genes of A. aegypti that overlap with mSDs are also associated with the same mSD copies in the one-to-one orthologous copies in C. quinquefasciatus and A. gambiae. Our results show that these mSDs are never retained in orthologous genes (data not shown), indicating the possibility of biased selection of mSD sequences of orthologous  genes. Such bias is most likely associated with purifying selection as such microsatellites in gene rich segmental duplications are known to be associated with such selection bias [22]. Furthermore, our analysis indicated that the mSDs within protein coding genes are mostly localized in the intron regions (data not shown). Hence, the lack of retention of mSDs in orthologous genes may also be due to higher rate of intron evolution than the coding sequences among the genes [32].
We also found evidence that mSDs are associated with gene duplications in specific species. The duplication of histone genes are clustered within a~113 kb region of chromosome 2L (2L:21421283-21534584) of D. melanogaster (Figure 9). In this region, twenty duplicated copies of a mSD was found wherein each copy (~3810 bp long) has (ATTT)n and (TC)n as termini sequences and harbors four histone genes (His1, His2A, His2B and His4). This pattern of duplication is often called 'duplication  shadowing' [33] as the duplicated sequences tend to cluster in the region along with an ancestral copy. In this region of the D. melanogaster genome, the locus corresponding to 2L: 21426344-21430203 is the ancestral duplication as evident from phylogenetic analysis ( Figure 9). As the duplications are anchored by same pair of microsatellites, it is likely that the ancestral copy might have expanded in that region by microsatellite mediated segmental duplications.
To further confirm if mSDs are associated with gene duplications, we identified the 'nearly identical paralogous' genes (NIPs, see Emrich et al. [31] for definition). We were able to find several NIPs in the C. quinquefasciatus and D. melanogaster genomes (Additional file 10) that were associated with mSDs. However, we didn't find any NIP associated with an mSD in A. aegypti, A. mellifera, A. gambiae and T. castaneum. Thus, if microsatellite mediated SDs have a role in gene duplication in insects; it is likely that such association is species-specific.
To determine if mSDs may have association with transposable elements (TEs) [34], we analyzed the annotated TEs to identify sequence duplications that are anchored by microsatellite pairs (Additional file 11). A list of the paired microsatellites associated with different TEs in D. melanogaster is provided in Additional file 12.
It shows that the total repertoire of mSDs associated with TEs is only a minor fraction of the total number of mSDs observed in Drosophila. This suggests that the mSDs are found primarily in TE-free and gene-free regions of the genome.

Discussion
In this study, we identified microsatellites that are repeated as pairs and investigated their association with segmentally duplicated sequences in insect genomes. We adopted a conservative approach to identify the repetitive microsatellite pairs in the genome by imposing the criterion that each pair has exactly the same intervening distance. However, we observed that, in some cases, the intervening distances are not exactly same but are similar (± 1 to 20 bp) among the microsatellite pairs. For example, the microsatellite pairs (ATTT)n and (TC)n are repeated 6 times wherein the intervening distance is exactly 3,807 bp (rMP family# 55, Additional file 2) compared to the other 14 duplications of the same SSR pairs but with intervening distance~3,810 bp ( Figure 9). The variation in intervening distances between the microsatellites may have resulted due to increase or decrease of repeat length of one or both of the microsatellites, possibly by slippage events during replications [35,36]. Slippage creates a loop in one of strands that gives rise to an insertion or a deletion in the subsequent replications depending upon if the loop is formed in the replicating strand or in the template strand respectively. This leads to an increase or decrease in repeat length of microsatellites. In most of the microsatellite pairs we identified, one of the microsatellites was variable in length while the length of the other microsatellite remained unchanged. It is known that sequence composition [35], imperfection in microsatellite motifs [14] and the local mutation rate of microsatellite loci [37] have roles in modulating the repeat length of the microsatellites that may account for variable intervening distances of paired microsatellites. Furthermore, differential selection of simple sequence coding repeats [10,38] may also account for the variation in distance between microsatellite pairs.
More than two microsatellites repeated together in the genome were also identified from our analysis.  21 ] is repeated three times in a tandem manner within the region 2L: 9475131-9483718 of D. melanogaster genome. However, such repeats containing more than two microsatellites were exceptionally low in frequency in the insect genomes (data not shown). On the other hand, repeats consisting of only two microsatellites are abundant in each species which was also observed by Kofler et al. [14]. Segmental duplications have been characterized in few organisms, mostly in the human and D. melanogaster genomes [39]. They are poorly studied in other species in spite of availability of draft genome sequences for many eukaryotes. Our study is a first effort in this direction to identify segmentally duplicated sequences from genome assemblies of different insects. In this study, the segmental duplications represent only a proportion of duplications where we find microsatellites at the sequence ends of the duplications. Although a comprehensive discovery of all the segmental duplications of these insects was not the aim of the present study, our results show that the repetitions of microsatellite pairs are associated with segmental duplications in insects but with extremely variable frequency. The A. aegypti and N. vitripennis genomes have more than one thousand mSDs whereas the T. castenium genome has only seven mSDs (Table 4) indicating that microsatellite anchored segmental duplications may be determined by species specific evolutionary processes.
Our results further showed that genomic regions with higher numbers of repetitive microsatellite pairs accumulate a greater number of segmental duplications than regions poor in paired microsatellites (Figure 7). This is a classic 'rich-gets-richer' mechanism where more segmental duplications tend to occur in regions that already have more duplicated sequences [21]. Such a mode of enrichment of SDs in specific chromosomal regions has relevance to 'duplication shadowing' effects in genome [33,40]. For example, duplication shadowing in the human genome contributes to~10 fold increased probability of sequence duplication in specific regions compared to their distribution in other regions [33]. We observed such a pattern of segmental duplications in chromosome 2L (2L: 21421283-21534584) of D. melanogaster where two (ATTT)n and (TC)n are repeated as sequence ends of each duplication (Figure 9). In this case, each SD contains four histone genes His1, His2A, His2B and His4. It was found that the entire~3810 bp sequence representing the segmental duplication maps to a single cDNA (accession # AY119274) suggesting that the duplicated sequence containing the four genes is expressed as a common primary transcript. It is possible that duplication shadowing of gene regions may be an evolutionary strategy to modulate expression of specific genes as evident in primates [33]. Moreover, Korbel et al. [41] also found that segmental duplications of larger sequences enclosing specific protein coding genes often contribute to the expansion of protein-coding gene families. Although the role of microsatellites in this process is not known, it has been found that microsatellites in the flanking sequences of genes may have a regulatory role in gene expression [5]. Moreover, simple sequence repeats in the coding region can influence translational selection of genes that can modulate expression level of those genes [10]. These reports indicate that microsatellite mediated segmental duplications may have an effect on expression of the genes when they are associated with segmental duplications in the genome.
The paired microsatellites identified from our investigation may be targets of non-homologous end joining (NHEJ), which is one of the mechanisms of segmental duplication [42]. Such processes are generally mediated by microhomologies (< 25 bp homology) at the ends of target sequences similar to the termini microsatellites of mSDs found in this study. Consistently, association of microsatellites has been indicated in genomic rearrangements [43] as well as segmental duplications [44]. Furthermore, it has been shown that microsatellites are enriched at breakpoints of SDs suggesting the possible role of microsatellite repeats in the genesis of SDs [20]. Hence, our results further corroborate that microsatellites, by repetition as pairs, are likely to have a role in the genesis of SDs in insect genomes.
It is also likely that mechanisms other than involving microsatellites have roles in segmental duplications. Nonallelic homologous recombination (NAHR) during meiosis using pre-existing repeat elements (such as Alu repeats) can also lead to segmental duplications [44]. Moreover, several factors such as length, orientation, degree of sequence similarity and the distance between the duplicated copies may lead to differential degrees of genomic rearrangements of sequences in genome [45]. It is possible that the genesis of segmental duplications may also be controlled by the same mechanisms that generate copy number variations (CNV) in genomes. CNVs are caused by different rearrangement events of sequences including deletions, duplications, inversions, and translocations [46]. However, Kim et al. found that only a minor portion (< 10%) of CNVs is associated with segmental duplications in the human genome suggesting independent mechanisms of genesis of SDs than that of CNVs.
Our data further suggests that duplications of paired microsatellites are localized mostly in the non-genic regions. In addition to that, the paired microsatellites in the genic regions are predominantly in the intron regions (data not shown). We also found several mSDs that are associated with different transposable elements (TEs) in the D. melanogaster genome (Additional file 12). Therefore, the role of microsatellites in intron evolution and retrotransposition events cannot be ruled out [32,47]. Given the role of transposition events in genome structure and function [48,49], it is likely that microsatellites are instrumental in extensive sequence transposition and duplication in the genome.