Secondary Logo

Journal Logo

Estimating the in-vivo HIV template switching and recombination rate

Cromer, Deborah; Grimm, Andrew J.; Schlub, Timothy E.; Mak, Johnson; Davenport, Miles P.

doi: 10.1097/QAD.0000000000000936

Background: HIV recombination has been estimated in vitro using a variety of approaches, and shows a high rate of template switching per reverse transcription event. In-vivo studies of recombination generally measure the accumulation of recombinant strains over time, and thus do not directly estimate a comparable template switching rate.

Method: To examine whether the estimated in-vitro template switching rate is representative of the rate that occurs during HIV infection in vivo, we adopted a novel approach, analysing single genome sequences from early founder viruses to study the in-vivo template switching rate in the env region of HIV.

Results: We estimated the in-vivo per cycle template switching rate to be between 0.5 and 1.5/1000 nt, or approximately 5–14 recombination events over the length of the HIV genome.

Conclusion: The in-vivo estimated template switching rate is close to the in-vitro estimated rate found in primary T lymphocytes but not macrophages, which is consistent with the majority of HIV infection occurring in T lymphocytes.

aComplex Systems in Biology Group, Centre for Vascular Research, University of New South Wales, Kensington

bInfection Analytics Program, Kirby Institute, University of New South Wales

cSchool of Public Health, Sydney University, Camperdown, New South Wales

dSchool of Medicine, Faculty of Health, Deakin University

eCSIRO Livestock Industries, Australian Animal Health Laboratory, Geelong, Victoria, Australia.

Correspondence to Miles P. Davenport, Infection Analytics Program, Kirby Institute, University of New South Wales, Kensington, NSW, Australia. Tel: +61 2 9385 0940; fax: +61 2 93851389; e-mail:

Received 23 March, 2015

Revised 9 October, 2015

Accepted 9 October, 2015

Supplemental digital content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's Website (

Back to Top | Article Outline


HIV mutation and recombination are essential to HIV viral evolution, immune evasion, and drug resistance in vivo[1–6]. Although mutation provides the building blocks for HIV diversification, retroviral recombination is a major ‘facilitator’ of the diversification process by bringing together mutations, yet there is currently no estimate on the rate of at which template switching occurs in vivo. HIV recombination has been studied extensively in vitro using both cell-free and cellular systems [7–13]. These studies have used either foreign fluorescent gene inserts or direct sequencing to identify recombination and reveal a high rate of recombination between both divergent (from different clades) or closely related genomes. In CD4+ T cells the in-vitro rate of template switching is approximately 3–15 recombination events per genome per reverse transcription cycle [7–13]. Some studies have also indicated that this rate is significantly higher in macrophages, though this is a subject of some debate [8,13,14]. These in-vitro infection studies have a number of constraints, including the use of nonviral gene reporter and marker HIV systems, cell lines for the production of virus or infection of cells, and a high level of activation of target cells for infection. Given the constraints of these in-vitro systems [7–13], it is currently unknown whether these in-vitro primary cell infection systems accurately reflect the recombination rate occurring during reverse transcription in vivo.

Studies of HIV recombination in vivo are significantly more complex because of a number of constraints that severely limit the estimation of template-switching rates. Firstly, although the process of template switching between the copackaged RNA strands occurs whenever reverse transcription is carried out; this switching only leads to observable recombination when it occurs in a heterozygous virion bearing two different templates (Fig. 1a and b). The generation of heterozygous virions in turn requires coinfection of cells with distinct viral genomes. The emergence of recombinant forms in vivo is thus highly influenced by cellular coinfection rates. Thus, in vivo there are two separate elements that contribute to observable recombination: the proportion of coinfection, which leads to heterozygous virions in which observable recombination can occur during reverse transcription (Fig. 2) and the template-switching rate per reverse transcription event. In vitro, it is possible to manipulate and measure the proportion of heterozygous virions directly, allowing mathematical compensation for the proportion of silent (homozygous) template switches to estimate the true rate of template switching [9]. In vivo, the rate of coinfection is not known, and estimates vary considerably from as low as 10% of HIV-infected T cells being coinfected [15–17] to as great as 80% [18], making it difficult to adjust for this factor. Additionally, since not all integrated HIV genomes can produce viruses, it is unclear what fraction of coinfected cells can produce the functional heterozygous virions required for recombination. Secondly, recombinant forms must also compete for resources with their founding parent virus. Selective advantages or fitness costs of recombinants will greatly influence the ability and the speed at which recombinant forms reach viral loads sufficient to be detected by sequencing.

Fig. 1

Fig. 1

Fig. 2

Fig. 2

For these reasons, in-vivo recombination studies often focus on the location of common recombination breakpoints [19–23], and the time frame until recombinant forms emerge in the viral quasi species [17]. The simplest opportunity for observing recombination in vivo occurs when a patient is infected with two different HIV clades. In this case, the evolution of recombinant forms has been relatively slow [15,17]. The observed rate of recombination in this scenario is a composite of the proportion of coinfected cells and heterozygous virions, as well as the template-switching rate during reverse transcription of these heterozygous virions. It is clear that one factor contributing to the slow rate of accumulation of recombinant virus in vivo is the low rate of coinfection.

We have previously developed a novel statistical technique for estimating the optimal-switching rate that would produce an observed number of recombination events during a single cycle of infection in vitro. This approach used a system of ‘marker’ viruses in which we have known mutations at certain positions in the genome. However, in vivo the identification of suitable ‘marker’ nucleotides identifying the different genomes and the estimation of the proportion of heterozygous virions is more difficult.

Here we present a novel method to separate the components of coinfection and per cycle template switching to directly estimate the rate of template switching in vivo using early viral sequences obtained by single genome analysis. Although the majority of HIV infections are thought to be initiated from a single founder virus, around 20% of infections are initiated by two or more closely related founder viruses [24]. In these circumstances we can use the single nucleotide differences between the different founder viruses as markers to identify the origins of different regions of the viral genome. Because sequence identity may influence rates of template switching [10,11,20,25,26], the rate of switching between these closely related founder viruses is likely more reflective of a ‘typical’ infection than previous studies investigating recombination in patients superinfected with multiple subtypes that have high sequence divergence. In addition, we avoid the problem of estimating the proportion of heterozygous virions by focusing only on the observed recombined sequences. That is, since we know that sequences with at least one recombination must have been derived from heterozygous virions, we effectively ask the question ‘given that this sequence was derived from a heterozygous virion and recombination occurred, what was the total number of template switches across our sequence?’ Therefore, taking into account the spacing of markers and observed number of recombination events in a sequence, we estimate upper and lower bounds for the single-cycle template-switching rate that would produce the observed data. Using this approach we produce the first direct estimate of the per reverse transcription template-switching rate in vivo.

Back to Top | Article Outline

Materials and methods

Identification of patients and sequences to be used in analysis

HIV-1 viral sequences for patients from published studies [27] and [28] were downloaded from National Center for Biotechnology Information. All patients, from [27] listed as containing recombinant sequences from fewer than 10 founder viruses, were included in our preliminary analysis. We also included some additional patients who were not included in [27] but were included in [28] and were listed as being in Fiebig stage 2–6 and as having infection from multiple founder viruses. All sequences were obtained using single genome amplification applied to HIV-1 env as described in [27,29]. In total 1194 sequences containing 3 355 643 nucleotides contributed to our analysis. The average sequence length considered was 2609 nucleotides.

Back to Top | Article Outline

Identification of founder sequences and marker nucleotides

Sequences from each patient were aligned using ClustalW and a single sequence with multiple identical copies of the sequence was identified as the first founder virus (Fig. 1a). For a sequence to be considered a founder, we required that at least two copies of it be observed. Positions at which a nucleotide (or two consecutive nucleotides) differed from the first founder in at least two sequences were identified as potential marker positions that could be used to infer where recombination between founders must have occurred.

Sequences identical to the first founder at all marker positions were classified as ‘founder 1’ sequences. Additional founder sequences were then identified as those sequences whose nucleotides at marker sites differed from founder 1, (but were identical between all sequences of the new founder). Identified founder sequences were required to differ from each other at a minimum of five marker positions.

Recombined sequences were identified as those sequences whose nucleotides at certain marker positions matched one founder, whereas at other marker positions matched a different founder. The makeup of such sequences could be explained using a combination of two or more founder viruses. As recombinants are only determined by differences at identified marker positions and cannot have any nucleotide at a marker position, but must have a nucleotide matching another founder at that marker position, the chances a sequence that underwent chance mutation being mistaken for a recombinant is very small. Given a mutation rate of 0.12/1000 nucleotides [9] and a total of 368 markers the probability of chance mutation being mistaken for recombination in any of our sequences is under 0.015. Figure 1 shows this method applied to one patient in particular, and a detailed worked example of this patient, as well as further details of our analysis, is given in the supplementary materials, We verified our detection of recombined sequences using both highlighter analyses and comparison with the output from Recco, software designed to automatically detect recombined sequences [30]. In all cases our detection of recombination either agreed with Recco, or else visual inspection indicated that our analysis provided a more accurate explanation for the observed sequence.

By taking into account the spacing of marker sites and the number of observed recombination events, we will be able to estimate bounds on the template-switching rate required to produce the observed recombination events.

Back to Top | Article Outline

Calculation of the in-vivo template-switching rate

To calculate the in-vivo template-switching rate, we consider only sequences in which recombination has been identified, as we can be sure these sequences were derived from a heterozygous virion. We determine the founder present at each marker position of the recombined sequence and the length of the interval between each marker. We then use maximum likelihood estimation to calculate the optimal template-switching rate that would have been required to generate the observed recombination events in the detected recombined sequences.

As previously described, our switching rate calculation accounts for the different lengths of intervals between marker sites (since a switch is more likely to occur over a longer interval) and allows for the possibility of multiple (unobserved) template switches to occur between these markers [9] [since simply because a recombination is not observed between two markers cannot discount the possibility that two (or any even number of) template switches occurred between these two markers, rendering the recombination undetectable, (Fig. 1c). This calculation is fully described in the supplementary materials,

Back to Top | Article Outline


To provide confidence intervals (CIs) on the estimated switching rates, we performed bootstrapping on the recombined sequences. We randomly selected an identical number of sequences (with replacement) as were used in the original calculation from the recombined sequences and used these randomly selected sequences to calculate the optimal template-switching rate. We performed 5000 iterations.

Back to Top | Article Outline

Accounting for multiple rounds of recombination

The method described above relies on the assumption that the recombination events observed in a sequence occurred during a single round of reverse transcription of a heterozygous virion. Though it does not require that recombination events from different sequences occurred at the same time, it does assume that once a recombined sequence was formed, it did not undergo a second round of observable recombination with another sequence. A graphical representation of this process is shown in Fig. 2. If recombination occurred over multiple cycles, then the actual per cycle template-switching rate would be lower than that estimated using the method above. Thus, using all suitable sequences and assuming that each observed recombination occurred during a single template switching cycle produces what we might regard as an ‘upper bound’ on the in-vivo per cycle recombination rate.

We, therefore, also derive a lower bound estimate for the in-vivo template-switching rate by considering only sequences in which a single recombination event is observed. In such sequences observable recombination must, by definition have occurred during a single round of reverse transcription.

Back to Top | Article Outline


Sequences used in analysis

Sequences from acutely infected patients early in HIV infection from [27] and [28] were analysed. There were 37 patients identified as having under 10 founder viruses and potentially containing a recombinant sequence. From these 37 patients we identified five patients in whom the sequences could be explained by exactly two different founder viruses (with more than one copy of each founder) with few mutations and containing at least one recombined sequence. These patients are listed in the top half of Table 1 and details of their sequences are shown in the supplementary materials, These five patients listed in Table 1 contained a total of 21 recombined sequences that contribute to the calculation of recombination rate.

Table 1

Table 1

Back to Top | Article Outline

Recombination rate

The method for estimating the recombination rate takes into account the unique spacing of markers on each sequence. In addition, it takes into account the possibility of template-switching events that do not result in an observed recombination event, that is where an even number of switches occurs in an interval between markers from the same founder or when more than one switch occurs in an interval between markers from different founders.

Using the methods described above and in the supplementary materials,, we estimate an upper bound on the template-switching rate that would explain the observed pattern of recombination seen in the recombined sequences listed in Table 1 to be 1.59/1000nt. Using bootstrapping, we estimated the 95% confidence interval (over 5000 runs) on the upper bound to be 1.24–1.99/1000nt. Restricting our analysis to only sequences in which we observed a single recombination, we estimate the lower bound of the recombination rate to be 0.50/1000nt (with CIs of 0.47–0.54/1000nt). These rates translate to between five and 14 switches per reverse transcription of the HIV genome. Owing to the relative symmetry of the bootstrap distributions, we assume very good coverage of these CIs.

Our published in-vitro estimate [9] and other published estimates of in-vitro recombination rates [7,8] lie within the upper and lower bounds estimated using this method.

Back to Top | Article Outline

Including patients with three founder viruses

When considering the 37 candidate patients from [27] and [28] we found an additional five patients in whom the detected sequences could be explained by three founder viruses, and in whom some sequences contained evidence of recombination between these founders. We next include these in our estimation of template-switching rate. These patients are shown in the bottom half of Table 1. Additional patients whose sequences were analysed but who were not included in our calculation of recombination rate are listed in Supplementary Table 1,

Detecting and analysing recombination in a sequence originating from a patient with three founders requires more detailed analysis than that described above, since not all marker sites contain information that identifies the founder virus present at the marker. That is, some marker positions may differentiate between founder 1 and founder 2, but be the same between founder 2 and founder 3, so a different pattern of markers exists between every pair of founders in a given patient. A detailed description of this process is given in the supplementary materials,

Including recombined sequences from the ‘three founder’ patients in our calculation results in an estimate of the upper bound on the template-switching rate of 1.54/1000nt (95% CI using bootstrapping of 5000 runs was 1.28–1.84/1000nt) and a lower bound of 0.58/1000nt (95% CI using bootstrapping of 5000 runs was 0.51–0.67/1000nt). This means that once we take into account the frequency and spacing of the observed recombination events we once again deduce that during each reverse transcription of the HIV genome, there are between 5 and 14 template switches between the strands that are copackaged into the HIV virion (again similar to published in-vitro estimates [7–9]). These template switches are only observed as recombination events when a heterozygous virion is transcribed and a marker nucleotide differs between the founders, so we can identify which founder the marker came from.

Back to Top | Article Outline


In this work, we set out to resolve the apparent discrepancy between the high template-switching rate estimated in vitro[7–9] and the observed low level of recombination in vivo (composite of both the actual template-switching rate and the low co-infection rate) [15]. As observable recombination is difficult in vivo compared with in vitro, and there is an unknown proportion of heterozygous virions present in vivo, the ‘true’ in-vivo template-switching rate is still a subject of debate [16,18]. We have successfully developed a new tool to bypass these constraints by focusing our studies on recombined sequences, which clearly originated from heterozygous virions, thereby enabling us to provide bounds on the in-vivo template-switching rate that would have produced the observed recombination events.

Our work has some perceived limitations. Firstly, if there is selection against the growth of highly recombined viruses in vivo, then this could have biased the sequences we observed and limited the number of recombined sequences for analysis. This limitation is in-part mitigated by the fact that we have chosen sequences from very early in infection when immune selection pressure is expected to be low. Secondly, when analyzing the patients with three founder viruses, where there was a choice in the potential makeup of a sequence, we chose the representation that would have led to the lowest estimated template-switching rate. However, we note that this does not seem to have biased results, as the estimates for the 2 founder and 3 founder groups were very similar.

We have provided both upper and lower bounds on the per cycle template-switching rate. These bounds are necessitated by the fact that we cannot confirm whether multiple observed recombinations occurred in a single round or over multiple rounds of infection. However, there are a number of factors that suggest that it is likely that most template switches occurred in a single round of infection. These are: if multiple rounds of recombination occurred, then in patients with more than two founder viruses we would expect to find some sequences that are recombinants of more than two founders (recombinant 1 in the red circle in Fig. 3b). However, we do not find any such sequences. In most patients the overall number of recombinant sequences was low, and these represented a small proportion of total sequences. If the frequency of having (at least) one recombination event is low, then the probability of two recombination events, in different rounds of infection, happening to the same sequence is even lower. If recombination events were occurring over multiple rounds of infection, then multiply recombined sequences should be more frequent in patients in later stages of infection than in patients in earlier stages. Thus, if such sequences were contributing to our estimate of template-switching rate, we would expect to estimate a higher rate in late stage patients. We found no significant difference in the estimated template-switching rates between patients in Fiebig stage 2 and Fiebig stage 5 patients (Fig. 4a). As the proportion of sequences with two recombination events should scale with the square of the proportion of sequences with one event, we would expect fewer ‘multiply recombined’ sequences in patients with a lower proportion of recombined sequences, and more in patients with a greater proportion of recombined sequences. Such an increase in the number of multiply recombined sequences should translate into a higher estimate of the template-switching rate. However, when we estimated the template-switching rate on subsets of our patients, selected based on the proportion of total sequences that were recombined, we observed no clear trend in the rate as the proportion of recombined sequences increased (Fig. 4b). When taken together, these multiple lines of evidence suggest that most recombined sequences did not form over multiple rounds of infection.

Fig. 3

Fig. 3

Fig. 4

Fig. 4

Finally, we have estimated the template-switching rate based on a region approximately 2600 nucleotides in the env region. We have recently shown that the rate of template switching varies across the HIV genome [31], and thus a gene-specific switching rate may be more appropriate. Owing to the limited availability of sequence data, it was not possible for us neither to compare the template-switching rate in this region with a rate in other regions nor to calculate the presence of recombination ‘hotspots’ over the env region.

The results of our study suggest that the recombination rate in vivo is 0.5–1.6 switching events per 1000 nucleotides, translating to ≈5–14 template switches over the entire HIV genome, and is likely to be closer to the upper bound provided. These bounds encompass previously estimated in-vitro recombination rates in CD4+ T cells both by us [9] and others [7,8]. Moreover, since both the calculated in-vitro recombination rate in CD4+ T cells and our estimated in-vivo switching rate are much lower than the in-vitro estimates of the recombination rate in macrophages [8], this suggests that in-vivo recombination events occur predominantly in CD4+ T cells, and not in macrophages. The high similarity between the in-vitro and in-vivo rates provides evidence that factors such as the genetic manipulation and in-vitro production of laboratory isolates of HIV and the sustained activation of T cells by mitogens do not significantly alter the rate of template switching of HIV during in-vitro infection. We have provided the first evidence that the published in-vitro estimates are reflective of the true per cycle recombination rate in vivo, and therefore that the results of in-vitro testing of drugs and therapies affecting HIV template-switching rates are likely to be transferable to in-vivo infections.

Back to Top | Article Outline


We would like to thank Diako Ebrahimi and Firoz Anwar for their assistance downloading the sequences from NCBI.

Back to Top | Article Outline

Financial support and sponsorship

This work is funded by grants from the Australian National Health and Medical Research Council and the Australian Research Council.

Back to Top | Article Outline

Conflicts of interest

J.M. is an ARC Future Fellow. M.P.D. is an NHMRC Senior Research Fellow. D.C., A.J.G., and M.P.D. conceived and designed experiments to address research question; D.C., A.J.G., T.E.S., and M.P.D. contributed to data analysis; D.C., T.E.S., J.M., and M.P.D. wrote the manuscript.

Back to Top | Article Outline


1. Rosenberg ES, Billingsley JM, Caliendo AM, Boswell SL, Sax PE, Kalams SA, et al. Vigorous HIV-1-specific CD4+ T cell responses associated with control of viremia. Science 1997; 278:1447–1450.
2. Evans DT, O’Connor DH, Jing P, Dzuris JL, Sidney J, da Silva J, et al. Virus-specific cytotoxic T-lymphocyte responses select for amino-acid variation in simian immunodeficiency virus Env and Nef. Nat Med 1999; 5:1270–1276.
3. Price DA, Goulder PJ, Klenerman P, Sewell AK, Easterbrook PJ, Troop M, et al. Positive selection of HIV-1 cytotoxic T lymphocyte escape variants during primary infection. Proc Natl Acad Sci U S A 1997; 94:1890–1895.
4. Borrow P, Lewicki H, Wei X, Horwitz MS, Peffer N, Meyers H, et al. Antiviral pressure exerted by HIV-1-specific cytotoxic T lymphocytes (CTLs) during primary infection demonstrated by rapid selection of CTL escape virus. Nat Med 1997; 3:205–211.
5. Meyerhans A, Cheynier R, Albert J, Seth M, Kwok S, Sninsky J, et al. Temporal fluctuations in HIV quasispecies in vivo are not reflected by sequential HIV isolations. Cell 1989; 58:901–910.
6. Goodenow M, Huet T, Saurin W, Kwok S, Sninsky J, Wain-Hobson S. HIV-1 isolates are rapidly evolving quasispecies: evidence for viral mixtures and preferred nucleotide substitutions. J Acquir Immune Defic Syndr 1989; 2:344–352.
7. Rhodes TD, Nikolaitchik O, Chen J, Powell D, Hu W-S. Genetic recombination of human immunodeficiency virus type 1 in one round of viral replication: effects of genetic distance, target cells, accessory genes, and lack of high negative interference in crossover events. J Virol 2005; 79:1666–1677.
8. Levy DN, Aldrovandi GM, Kutsch O, Shaw GM. Dynamics of HIV-1 recombination in its natural target cells. Proc Natl Acad Sci U S A 2004; 101:4204–4209.
9. Schlub T, Smyth R, Grimm A, Mak J, Davenport M. Accurately measuring recombination between closely related HIV-1 genomes. PLoS Comput Biol 2010; 6:e1000766.
10. Zhuang J, Jetzt AE, Sun G, Yu H, Klarmann G, Ron Y, et al. Human immunodeficiency virus type 1 recombination: rate, fidelity, and putative hot spots. J Virol 2002; 76:11273–11282.
11. Jetzt AE, Yu H, Klarmann GJ, Ron Y, Preston BD, Dougherty JP. High rate of recombination throughout the human immunodeficiency virus type 1 genome. J Virol 2000; 74:1234–1240.
12. Chen J, Powell D, Hu W-S. High frequency of genetic recombination is a common feature of primate lentivirus replication. J Virol 2006; 80:9651–9658.
13. Chen J, Rhodes TD, Hu W-S. Comparison of the genetic recombination rates of human immunodeficiency virus type 1 in macrophages and T cells. J Virol 2005; 79:9337–9340.
14. Nguyen LA, Kim D-H, Daly MB, Allan KC, Kim B. Host SAMHD1 protein promotes HIV-1 recombination in macrophages. J Biol Chem 2014; 289:2489–2496.
15. Neher R, Leitner T. Recombination rate and selection strength in HIV intra-patient evolution. PLoS Comput Biol 2010; 6:e1000660.
16. Josefsson L, King MS, Makitalo B, Brännström J, Shao W, Maldarelli F, et al. Majority of CD4+ T cells from peripheral blood of HIV-1-infected individuals contain only one HIV DNA molecule. Proc Natl Acad Sci U S A 2011; 108:11199–11204.
17. Batorsky R, Kearney MF, Palmer SE, Maldarelli F, Rouzine IM, Coffin JM. Estimate of effective recombination rate and average selection coefficient for HIV in chronic infection. Proc Natl Acad Sci U S A 2011; 108:5661–5666.
18. Jung A, Maier R, Vartanian J-P, Bocharov G, Jung V, Fischer U, et al. Recombination: multiply infected spleen cells in HIV patients. Nature 2002; 418:144.
19. Baird HA, Galetto R, Gao Y, Simon-Loriere E, Abreha M, Archer J, et al. Sequence determinants of breakpoint location during HIV-1 intersubtype recombination. Nucleic Acids Res 2006; 34:5203–5216.
20. Baird HA, Gao Y, Galetto R, Lalonde M, Anthony RM, Giacomoni V, et al. Influence of sequence identity and unique breakpoints on the frequency of intersubtype HIV-1 recombination. Retrovirology 2006; 3:91.
21. Archer J, Pinney JW, Fan J, Simon-Loriere E, Arts EJ, Negroni M, et al. Identifying the important HIV-1 recombination breakpoints. PLoS Comput Biol 2008; 4:e1000178.
22. Zhang M, Foley B, Schultz A-K, Macke JP, Bulla I, Stanke M, et al. The role of recombination in the emergence of a complex and dynamic HIV epidemic. Retrovirology 2010; 7:25.
23. Simon-Loriere E, Galetto R, Hamoudi M, Archer J, Lefeuvre P, Martin DP, et al. Molecular mechanisms of recombination restriction in the envelope gene of the human immunodeficiency virus. PLoS Pathog 2009; 5:e1000418.
24. Keele BF, Giorgi EE, Salazar-Gonzalez JF, Decker JM, Pham KT, Salazar MG, et al. Identification and characterization of transmitted and early founder virus envelopes in primary HIV-1 infection. Proc Natl Acad Sci U S A 2008; 105:7552–7557.
25. Chin MPS, Rhodes TD, Chen J, Fu W, Hu W-S. Identification of a major restriction in HIV-1 intersubtype recombination. Proc Natl Acad Sci U S A 2005; 102:9002–9007.
26. Iglesias-Sánchez MJ, López-Galíndez C. Analysis, quantification, and evolutionary consequences of HIV-1 in vitro recombination. Virology 2002; 304:392–402.
27. Li H, Bar KJ, Wang S, Decker JM, Chen Y, Sun C, et al. High multiplicity infection by HIV-1 in men who have sex with men. PLoS Pathog 2010; 6:e1000890.
28. Gnanakaran S, Bhattacharya T, Daniels M, Keele BF, Hraber PT, Lapedes AS, et al. Recurrent signature patterns in HIV-1 B clade envelope glycoproteins associated with either early or chronic infections. PLoS Pathog 2011; 7:e1002209.
29. Salazar-Gonzalez JF, Bailes E, Pham KT, Salazar MG, Guffey MB, Keele BF, et al. Deciphering human immunodeficiency virus type 1 transmission and early envelope diversification by single-genome amplification and sequencing. J Virol 2008; 82:3952–3970.
30. Maydt J, Lengauer T. Recco: recombination analysis using cost optimization. Bioinformatics 2006; 22:1064–1071.
31. Smyth RP, Schlub TE, Grimm AJ, Waugh C, Ellenberg P, Chopra A, et al. Identifying recombination hot spots in the HIV-1 genome. J Virol 2014; 88:2891–2902.

HIV; HIV evolution; in vivo; mathematical modelling; recombination rate; reverse transcription; template switching

Supplemental Digital Content

Back to Top | Article Outline
Copyright © 2016 Wolters Kluwer Health, Inc.