Abstract
Background
HIV and HCV infections have become the leading global publichealth threats. Even more remarkable, HIVHCV coinfection is rapidly emerging as a major cause of morbidity and mortality throughout the world, due to the common rapid mutation characteristics of the two viruses as well as their similar complex influence to immunology system. Although considerable progresses have been made on the study of the infection of HIV and HCV respectively, few researches have been conducted on the investigation of the molecular mechanism of their coinfection and designing of the multitarget coinhibitors for the two viruses simultaneously.
Results
In our study, a multitarget Quantitative StructureActivity Relationship (QSAR) study of the inhibitors for HIVHCV coinfection were addressed with an insilico machine learning technique, i.e. multitask learning, to help to guide the coinhibitor design. Firstly, an integrated dataset with 3 HIV inhibitor subsets targeted on protease, integrase and reverse transcriptase respectively, together with another 6 subsets of 2 HCV inhibitors targeted on NS3 serine protease and NS5B polymerase respectively were compiled. Secondly, an efficient multitarget QSAR modelling of HIVHCV coinhibitors was performed by applying an accelerated gradient method based multitask learning on the whole 9 datasets. Furthermore, by solving the L1infinity regularized optimization, the Druglike index features for compound description were ranked according to their joint importance in multitarget QSAR modelling of HIV and HCV. Finally, a drug structureactivity simulation for investigating the relationships between compound structures and binding affinities was presented based on our multiple target analysis, which is then providing several novel clues for the design of multitarget HIVHCV coinhibitors with increasing likelihood of successful therapies on HIV, HCV and HIVHCV coinfection.
Conclusions
The framework presented in our study provided an efficient way to identify and design inhibitors that simultaneously and selectively bind to multiple targets from multiple viruses with high affinity, and will definitely shed new lights on the future work of inhibitor synthesis for multitarget HIV, HCV, and HIVHCV coinfection treatments.
Background
Human immunodeficiency virus (HIV1) is the cause of acquired immunodeficiency syndrome (AIDS) which has infected more than 60 million people around the world [1,2]. Meanwhile, Hepatitis C virus (HCV), which is served as a serious cause of chronic liver disease, has infected 150200 million people worldwide [3]. Nowadays HIV and HCV infections have become global publichealth threats. Even more remarkable, HIVHCV coinfection is rapidly emerging as a major cause of morbidity and mortality throughout the world, since that both of the viruses share the same routes of transmission [3,4]. It is shown that infection with the HCV is the most common coinfection in people with HIV, and hepatitis C is categorized as an HIVrelated opportunistic illness. Complications related to HIVHCV coinfection are becoming an increasingly important medical issue [4].
The current strategies for developing HIV/HCV antiviral agents depend essentially on disrupting the replication of the 2 viruses, and various inhibitors have been designed to target and block the functions of the enzymes necessary in the replication cycle of HIV/HCV. Among them, HIV inhibitors commonly target on protease, integrase and reverse transcriptase (RT), while HCV inhibitors target on NS5B polymerase and NS3 serine protease [518]. These inhibitors have been considered as attractive targets for therapeutic intervention in HIV/HCV infected patients.
For HIV and HCV therapy, single antiretroviral drug, alone or in simply combination with each other, is no longer recommended for clinical use owing to (1) the complicated infection mechanism of these two viruses; (2) the severe side effects of the joint using and (3) the rapid emergence of drugresistant strains after initiation of therapy. Hence, drugs targeting on different targets with high therapeutic and reduced side effects are expected to be more effective at suppressing viral growth. For HIV, The multitarget antiretroviral drugs can succeed in inhibiting several HIV proteins simultaneously and efficiently. There has existed several pioneering work in multitarget drug discovery for HIV infection, such as the multitarget antiretroviral drug Cosalane [13], which was developed to inhibit several HIV1 proteins simultaneously. Compared to HIV, the multiple target HCV drug treatment is still in its infancy. Nevertheless, the combination use of singletarget HCV drugs has become a new chance in this field, such as the combination using of NS5B polymerase inhibitor (GS9190) and NS3 protease inhibitor (GS9256), which were shown to be safe, welltolerated and show dose dependant antiviral activity [19,20].
Since for both HIV and HCV the smallmolecule compounds used to design the drugs are needed to be assayed in vitro and in vivo, the popular insilico Quantitative StructureActivity Relationship (QSAR) modelling is applied extensively in HIV/HCV inhibitor studies due to its charming "blackbox" characteristics as well as its well prediction ability. Normally the QSAR modelling can be viewed as a computational technique to elucidate a quantitative correlation between chemical structure and biological activity [21]. Recently, considerable QSAR studies have been made for HIV/HCV inhibitors studies [518]. However, these studies were mainly focused on specific types of targets or specific diseases individually. Few studies have been performed on the multitarget HIVHCV coinfection QSAR modelling. Although the ways in which coinfection with HIV and HCV affect the body are still poorly understood, it has been indicated that both HIV1 Protease and HCV NS3 Protease are responsible for cleaving the viral polyproteins during the course of their action to produce the individual proteins of the mature viruses. Similarly, HIV1 reverse transcriptase and HCV NS5B can be affected by either nucleoside inhibitor that terminates nucleic acid synthesis or nonnucleoside inhibitor that impairs enzymatic function [22,23]. All these evidences have indicated that it is possible to design certain inhibitors that aim at both HIV targets and HCV targets simultaneously. From this point of view, multitarget coinfection QSAR modelling for HIV and HCV is attractive and promising, due to that it is easy to achieve and expected to provide useful clues on how to synthesize such coinhibitors with improved affinities.
In our previous study, we presented a multitarget QSAR modelling on HIV1 inhibitors individually [31]. In this study we desire to extend this model to investigate the multitarget QSAR modelling of HIV and HCV jointly and simultaneously, and aim at providing useful clues on the design of HIVHCV coinhibitors. The QSAR modelling of HIVHCV coinfection inhibitors (coinhibitors for short) was addressed by applying an efficient accelerated gradient method based multitask learning (MTL) model provided by us formerly in machine learning community [24]. QSAR studies were performed on 9 datasets of HIV and HCV inhibitors. By using our MTL framework, the correlations among different set of inhibitors were utilized and an efficient multitarget QSAR modelling of HIVHCV coinhibitors was obtained. According to the importance of each descriptor in QSAR model, the Druglike index (DL) features [25] for inhibitor description were ranked, and a drug structureactivity simulation were performed to investigate the relationships between compound structures and binding affinities based on the ranked molecule descriptors.
Methods
A Dataset
Our integrated dataset contains 3 kinds of HIV target subsets and 6 kinds of HCV target subsets, which were compiled from a thoroughly literature reviewing, consisting of inhibitors with their binding affinities on HIV protease, integrase and reverse RT, as well as HCV NS3 and NS5B respectively. This data provided the first time a comprehensive data source for multitarget HIVHCV coinfection QSAR study. In our study, these inhibitors are correspondingly referred as (1) protease inhibitors, which prevent HIV from processing and packaging new virulent viral particles, (2) integrase inhibitors, which inhibit the proviral DNA to insert into the host cell genome, (3) nonnucleoside reverse transcriptase inhibitors (NNRTI), which inhibit the virus by preventing the transcribing of its genomic DNA into proviral DNA for incorporation into the host cell DNA, (4) NS3 serine protease inhibitors, which prevent polyprotein processing and restore the hepatocytes innate antiviral response, and (5) NS5B polymerase inhibitors, which prohibit the synthesis of RNA strands of HCV. All the enzymes affected by these inhibitors have been reported as the most important targets for chemotherapeutic agents against the diseases caused by HIV/HCV. General descriptions of inhibitors for these targets were listed in Table 1.
Table 1. Dataset descriptions.
Similar to our previous study [31], the inhibitors were represented with 2 kinds of feature spaces referring to 32dimensional General Descriptor (GD) features and 28dimensional Druglike index (DL) features. Although there are numerous types of descriptors to describe a chemical compound, none of a set of descriptors can guarantee to behave overwhelming better than others. Therefore, the widely applicable set of descriptors, i.e., the GD [25] was selected, together with the DL descriptor [26,27] as a complement.
Detailed biological meaning of GD and DL descriptors can be referred in our previous work [31]. It should be noted that: (1) normally, general descriptors characterize physical prosperities of compounds, while druglike index descriptors characterize simple topological indices of compounds. These two kinds of descriptors are expected to present a comprehensive description of the compounds from the views of their intrinsic characteristics as well as their druglike properties. (2) The GD descriptor is generated in a hybridized way thus its current features haven't kept their original means for compound structure description. Therefore it cannot be biologically explained easily. On the other side, DL holds its original meanings, thus will be applied in our following feature ranking and explanations.
It was shown in Table 1 that the inhibitor activity of the molecules were measured with EC_{50}, IC_{50 }or PK_{i}, which are the most commonly used measurements of the compound inhibitions [28]. EC_{50 }(half maximal effective concentration) refers to the concentration of a drug, antibody or toxicant which induces a response halfway between the baseline and maximum after some specified exposure time, while IC_{50 }(half maximal inhibitory concentration) refers to the concentration it needed to inhibit a given biological process or component of a process by half. IC_{50 }can be converted to the PK_{i }measurement by the ChengPrusoff equation:
Where [L] is the concentration of free radio ligand used and K_{D }is its equilibrium dissociation constant for the receptor [29].
It should be noted that the QSAR data were provided by different research groups under different platforms/protocols with different activity measurements. Normally QSAR modeling achieved by such single target data is often not reliable due to the insufficiency of samples. However, since we want to investigate the multitarget QSAR relationship of the HIVHCV coinfection, these data can be integrated in an elegant multitarget QSAR model taking the advantages of the multitask learning [30], which would expect to exploit the possible synergies between different datasets and obtain a better QSAR model to guide the synthesis of certain inhibitors with enhanced activities for HIV and HCV simultaneously. Details will be shown in the following.
B Methodology
Computational framework for multitarget modelling
The general computational pipeline for our study was presented in Figure 1. In our previous study we have presented the first time a multitask learning algorithm for crossplatform siRNA efficacy prediction [30], and also utilized such MTLbased model for featureselection in HIV1 QSAR modeling [31]. It has been proven to be more effective than learning each QSAR modeling on single target independently [31]. Some latent commonalities across tasks can be exploited through MTL, which is expected to boost the learning performance of each single task.
Figure 1. The general computational framework for multitarget QSAR modelling of HIVHCV coinhibitors.
In current study, a novel accelerated gradient descent algorithm based MTL model was performed for multitarget QSAR modeling on our integrated datasets simultaneously. Our inhouse experiments indicated that this MTL model is more efficient than our formerly adopted one for multitarget QSAR modeling [31] and it is scaled up well for large scale QSAR modeling in both convergence speed and learning accuracy. A joint L1infinity regularization based feature selection procedure was performed on the DL feature space to reveal the most common features across multitarget HIVHCV coinfection QSAR modeling. Based on such model, a drug structureactivity simulation for investigating the relationships between compound structures and binding affinities was further presented to validate our selected important features for efficient coinhibitor synthesis and design.
Multitask learning for QSAR modelling of HIVHCV coinhibitors
Multitask learning has been developed in machine learning research to situations where multiple related learning tasks are accomplished together. It has been proven to be more effective than learning each task independently when there are explicit or hidden interrelationship among the tasks that can be exploited. The intuition underlying the framework is that the multiple related tasks can benefit each other by sharing the data and features across the tasks, which can often boost the learning performance of each single task [30]. Also it provides an efficient mechanism for crosstask feature selection, thus could uncover the common dominate features for all the tasks simultaneously. Such computational ability is inherently suitable for our multitarget QSAR modeling, in which each single QSAR model could be viewed as a task and the leading features for synthesizing coinhibitors with improved activity will be identified under such schema.
It should be noted that the QSAR modelling is the process by which chemical structure is quantitatively correlated with a welldefined process, such as biological activity or chemical reactivity. And this procedure is generally formulated as a regression model [32] to predict the compound activity based on a given set of molecule descriptors. Although various statistical and machine learning methods have been proposed in the last few years for QSAR modeling [32], few studies have been tried in the multitarget QSAR scenario. In our study the multitarget QSAR modeling will be elegantly formulated as a multitask regression framework to reveal useful clues for multitarget drug screening and synthesizing for HIVHCV coinfections.
Basically in our multitarget modeling, assuming the datasets contains N tuples, z_{i }= (x_{i}, y_{i}, k_{i}) for i = {1...N}, where x_{i }∈ R^{d }is the inhibitor descriptor and k_{i }∈ {1...M} is the indicator specifying to which of the M task the example (x_{i}, y_{i}) corresponds to. A critical issue in this case is to learn a set of sparse functions across these tasks for activity regression. Specifically, the features will be represented as GD or DL. y_{i }is the corresponding inhibitor binding affinity. Our goal is to predict the target binding affinity from a set of compound with known affinity by learning M linear regressions of the form . In our study the following square loss function is adopted:
where z = (x, y, k), W =[w_{1}, w_{2}, ..., w_{M}] ∈ R^{d × M }and W^{j }be the jth row of W.
In sparse MTL for features selection, we enforce the joint sparsity across different tasks by adding the l_{1,∞ }norm of the matrix W to the square loss function, which leads to only a few nonzero rows of W, and thus the corresponding features will be used for prediction. In short, the optimization target function is defined as F(W), and we want to minimized the following function as:
where
and
The l_{1,∞ }norm of the matrix W is defined as:
The first term in Equation (3) is the average of the empirical error across the tasks. The second term is the L1infinity regularization term that works on feature selection task in MTL, which can yield joint sparsity on both the feature level and task level and can lead to a more sparse solution [24].
As the main difficulty for solving the l_{1,∞ }regularized formulation in formation (6) lies in the nonsmooth property of the l_{1,∞ }regularizer, we present an accelerated gradient descent algorithm with the convergence rate O (1/t^{2}) by a variation of Nesterov's method calling a blackbox oracle in the projection step at each iteration [24]. By exploiting the structure of the l_{1,∞ }ball, we find the blackbox oracle can be efficiently solved by a simple sorting procedure. Compared with Nesterov's algorithm, our method is suitable for largescale multitask learning problem since it only utilizes the first order information and is very easy to implement. Experiment results in our previous study have shown that our method significantly outperforms the most stateoftheart methods in both convergence speed and learning accuracy [25].
Details of the accelerated gradient descent algorithm were presented in Algorithm 1. The generalized gradient update step was defined as following:
where ·_{F }denotes the Frobenius norm and 〈A, B〉 = Tr(A^{T }B) denotes the matrix inner product.
Algorithm 1: Accelerated Gradient Algorithm
Initialization: L_{0 }> 0, η > 1, W_{0 }∈ R^{d × M}, V_{0 }= W_{0 }and a_{0 }= 1.
Iterate for t = 0, 1, 2,... until convergence of W_{t}:
1) Set L = L_{t}
2) While F(q_{L}(V_{t})) > Q_{L}(q_{L}(V_{t}), V_{t})
L = η L
3) Set L_{t+1 }= L and compute
In addition, we suggest a lookahead stopping criterion for Algorithm 1 by firstly fixing a step size h and in each iteration t calculating the following ratio:
We stop the procedure when κ ≤ τ where τ is a prefixed constant.
Then, we discuss how to solve the generalized gradient update efficiently. Rewrite formulation (7), we obtain that:
For the sake of simplicity, we denote as V and as . For equation (9) the following form was taken:
where W^{i}, V^{i }denotes the i th row of the matrix W, V respectively. Therefore, (10) can be decomposed into d separate subproblems of dimension M.
For each subproblem:
Since the conjugate of a quadratic function is still a quadratic function and the conjugate of the norm is the barrier function, the dual of (11) takes the following form:
and the vector of dual variables α satisfies the relation α = v  w. Equation (12) can be solved by an efficient projection onto the ball l_{1 }according to [33]. With the primal dual relationship, we present Algorithm 2 for solving (11):
Algorithm 2: Algorithm for projection onto the l_{∞ }ball
Input: A vector v ∈ R^{M }and a scalar
1) If v_{1 }, set w = 0. Return.
2) Let u_{i }be the absolute value of v_{i}, i.e. u_{i }= v_{i}. Sort vector u in the decreasing order: u_{1 }≥ u_{2 }≥ ... ≥ u_{M}
3) Find
Output:
Feature selection across multiple tasks
It has been demonstrated that the coefficient among different tasks will achieve zeros simultaneously by using the l_{1,∞ }norm regularization with the "grouping" effect [24]. So the l_{1,∞ }regularization can provide an efficient way to evaluate the joint feature importance in HIVHCV coinhibitor design across multiple target enzymes. The basic idea behind this joint feature selection procedure is that the parameter W outputted from the algorithm for projection onto the l_{1,∞ }ball would present a ranking of the features based on their joint importance across multiple tasks. Formally, based on the parameter W derived from Algorithm 2, we can obtain β_{i }according to the following equation
The value of β_{i }indicates the weight of the corresponding feature, which gives us a quantitative way to evaluate the importance of various features for HIVHCV coinhibitor design and synthesize.
Domain of applicability of the model
The domain of application (DOA) is an important issue for QSAR model which is used for estimating the reliability in the prediction of a new compound [34]. Those molecules which fall out the domain may lead to unreliable predictions. Extent of extrapolation is a simple method to define the DOA[35]. It is based on the value of leverage h_{i }define in equation (14) for each chemical molecule:
Where X_{i }is the rowvector descriptor of the query compound, X_{i }is the n × k matrix containing k descriptor values and n training samples. The superscript T refers to the transpose of the matrix or vector. Generally, the warning leverage h* is fixed at 3 k/n, where n is the number of training compounds, and k is the number of descriptor. When a leverage is greater than the warning leverage h*, the predicted activity is the result of substantial extrapolation of the model and, therefore, it may not be reliable.
Based on the definition of leverage, Williams plot was used in our study to visualize the DOA of the QSAR model [35]. The Williams plot plots the standardized crossvalidated residuals (RES) versus leverage values (h), and can be used to obtain an immediate and simple graphical detection of both the response outliers (Y outliers) and the structurally influential chemicals (X outliers) of a model. Generally, the points with their values of Y axis fall outside the 3σ line (σ is the standard residuals unit of the compounds) can be referred as the Y outliers, while the points with their values of X axis fall outside the warning leverage h* line can be referred as the X outliers.
Multitarget HIVHCV coinhibitor design based on drug structureactivity prediction
After the feature ranking together with the examination of domain of application for multiple HIVHCV drug targets QSAR modelling, a drug structureactivity prediction [27] was performed for the analysis of the multiple drug data. The goal of this study is two folds: (1) It is used to computationally validate the ranking result by our multitask feature selection, and (2) It provides several useful modification strategies for further HIVHCV coinhibitor design.
Our prediction pipeline is carefully designed as shown in Figure 2. To be brief, the whole procedure is achieved by the following steps: First, All the compounds used for our multitarget QSAR study are gathered as the input, to generate their common scaffolds. If more than one scaffold is presented, they are topologically aligned to produce a common numbering system. Second, assigning the scaffolds by enumerating all possible substructure matches, then minimizing a pairwise energy term which leads to the lowest possible diversity of implied Rgroup substituents, which is expected to provide a set of analogous substitution points. An example in this case (Figure 3), the positions indicated by R1 for each molecule are, for this example, considered analogous substitution points, as are the positions indicated by R2. These positions have common meaning for the molecules, regardless of which of these two scaffolds they are based on. Finally, the list of hypothetical molecules, constructed from available ones, is generated by enumerating all of the input molecules, and performing singlepoint mutations at each of the substituent positions, with each of the Rgroups that have been observed in the analogous position for some other molecule in the input dataset. The unique list of chimeric molecules is then scored according to an estimate of probability, scaled and balanced to match the distribution of activities found in the input set. The scores are scaled such that a value of 0 indicates that the hypothetical molecule is as likely to be active as an average molecule in the input set, while positive values are more likely [27].
Figure 2. Computational pipeline for structureactivity simulation and hypothetical molecule generation.
Figure 3. An example of analogous substitution points. The positions indicated by R1 for each molecule are considered analogous substitution points as the positions indicated by R2.
Based on this computational prediction pipeline, we will identify what is the most efficient compound modification strategy to improve the molecule affinity targeting on multiple HIVHCV enzymes. Also, these identified strategies will be further explained by our joint feature ranking obtained under the multitarget QSAR paradigm.
Results and Discussions
Formulated as a multitask regression problem, the QSAR modelling of HIVHCV coinhibitors was performed based on the accelerated gradient descent based sparse multitask learning. Root mean squared error (RMSE) and squared correlation coefficient (R^{2}) were adopted as the performance evaluation for testing results. The definitions of these statistical parameters are provided as followed:
Root mean squared error (RMSE):
where n is the number of predicted drug molecules
, is the difference between the observed molecule affinity data and the fitted model
y_{i }is the observed molecule affinity
is the predicted molecule affinity
Squared correlation coefficient (R^{2}):
where P^{avg }is the average value of over the n predicted molecule affinities.
Sparse MTL was trained jointly on 9 targets datasets as indicated in Table 1 with 10%, 30%, 50%, 70% and 90% of the whole data respectively. The trained model was then applied for the affinity prediction of remaining data. We run each experiment 10 times and output the average RMSE and R^{2}. For each target, the output were calculated based on single task learning (denoted as STL in the figure), multiple task learning on HCV or HIV (denoted as HCV/HIV in the figure) and multiple task learning on all the inhibitors (denoted as MTLALL in the figure). The testing results were summarized in Figure 4, 5, 6 and 7, denoting the scenarios of representation of compounds with GD and DL respectively. The goal of this test was to compare the performance of MTL and singletask learning (STL) in the QSAR modelling of HIV/HCV inhibitors.
Figure 4. RMSE comparison of QSAR modelling based on accelerated gradient descent based MTL with STL on 9 targets, described in GD representation.
Figure 5. RMSE comparison of QSAR modelling based on accelerated gradient descent based MTL with STL on 9 targets, described in DL representation.
Figure 6. R^{2 }comparison of QSAR modelling based on accelerated gradient descent based MTL with STL on 9 targets, described in GD representation.
Figure 7. R^{2 }comparison of QSAR modelling based on accelerated gradient descent based MTL with STL on 9 targets, described in DL representation.
It is obvious that for both GD and DI feature space, using multitask learning for QSAR modelling is superior to singletask learning on most target datasets, with the evaluation of RMSE and R^{2 }and significant statistical confidence (data not shown). And the average correlation coefficient for data prediction under MTL is up to 0.6~0.7, which is a wellaccepted QSAR results. Such results proved that multitask learning can discover the latent commonalities across different types of inhibitors and take advantage of the synergy among multiple tasks when the label data on each single task are insufficient. These results also indicated that multitask learning provides an effective way to boost the learning performance of each single task by exploiting the available synergy between them, thus served as an efficient paradigm for multitarget QSAR modelling.
In order to define the domain of applicability of our QSAR model, the Williams plots were drew for each QSAR model, based on MTL and STL with GD and DI feature space respectively. For each dataset, 65% of the whole data was chosen as the train data and the remains were used as testing. The plots were summarized from Figure 8, 9, 10 and 11. Each plot contains the compounds represented as the training set (purple squares) and test set (red squares). It can be clearly seen that most of the compounds fall into their corresponding application domain, which indicated that for both MTL and STL the chemical molecules are following a welldefined domain of applicability. It is worth noting that for most of those molecules following outside the DOA, their values of Y axis are generally fall inside the 3σ line, which indicated that, to a large extent, the predictions of our model for the outliers may still be reliable. In addition, very few compounds in several models indeed fall outside the 3σ line, however, they are close to the line thus can be retained.
Figure 8. Williams plots of 9 QSAR models based on MTL with Druglike Descriptor.
Figure 9. Williams plots of 9 QSAR models based on STL with Druglike Descriptor.
Figure 10. Williams plots of 9 QSAR models based on MTL with General Descriptor.
Figure 11. Williams plots of 9 QSAR models based on STL with General Descriptor.
After building the QSAR model, the weight of the DL features for MTL on 3 HIV datasets, 6 HCV datasets and all 9 datasets respectively were calculated, Sparse MTL in this case was trained with 50% of the data for each task and tested with the remaining data. The features ranking were showed in Figure 8, 9 and 10. It should be noted that the GD feature space will not be adopted for feature ranking due to its indirectly mapping of biological meanings.
Table 2 were concluded from Figure 12, 13 and 14 to provide a joint feature ranking of the DL compound descriptors. Top five features were selected to guide our structureactivity prediction as implemented with structureactivity report (SAReport) in MOE [27]. These features are # of nonH, # of 2degree acyclic atoms, degree of cyclization, # of nonH polar bonds and # of rotatable bonds.
Table 2. A joint feature ranking of the DL compound descriptors.
Figure 12. Weights of Druglike features of HIV inhibitors.
Figure 13. Weights of Druglike features of HCV inhibitors.
Figure 14. Weights of Druglike features of HIVHCV inhibitors.
SAReport was applied to present a direct instruction on how to modify the structure of a compound to make it to be a better multitarget coinhibitor of HIV and HCV. The top structures were selected with their ranking of affinity improvements according to various modification mechanisms. As indicated from Table 3, 4, 5, 6 and 7, the modifications of compound based on these structure features are expected to be a feasible way for efficient multitarget coinhibitor design. Generally, the following useful modification strategies were obtained, i.e., (1) Improving the # of nonH; (2) Improving the # of 2degree acyclic atoms; (3) Improving the degree of cyclization; (4) Improving the # of nonH polar bonds and (5) Improving the # of rotatable bonds. These modifications will improve the binding affinities of HIV and HCV inhibitors respectively. Such common mechanism for improving the inhibitor's affinity on different antiHIV or antiHCV targets is consistent with our ranking of joint feature selection obtained by multitarget coinhibitors QSAR modelling.
Table 3. More # of nonH, # of nonH polar bonds and # of rotatable bonds could increase the potency of HIV Protease Inhibitors.
Table 4. More # of 2degree cyclic atoms, degree of cyclization and # of nonH polar bonds could increase the potency of HIV Reverse Transcriptase Inhibitors.
Table 5. More # of nonH, # of 2degree cyclic atoms, degree of cyclization and # of nonH polar bonds could increase the potency of HIV Integrase Inhibitors.
Table 6. 5 More # of nonH, # of nonH polar bonds and # of rotatable bonds could increase the potency of HCV NS5B Inhibitors.
Table 7. More # of nonH, # of 2degree cyclic atoms, degree of cyclization and # of nonH polar bonds could increase the potency of HCV NS3 Inhibitors.
It should be noted that these features are well be commonly important for the multiple scaffolds with each inhibiting an individual target derived from our MTL model, and they can be necessarily integrated together to guide the synthesis of a single scaffold against multiple targets, which guarantee that one individual compound could hold its coinhibitor ability for both virus targets.
Conclusions
A Multitarget computational screening of HIVHCV coinhibitors with a MTL paradigm was carried out in our study. Compared to our previous work [31], it is improved mainly in two aspects: (1) It integrated both HIV and HCV data sources to enhance significantly the identification of lead inhibitors for HIVHCV coinhibitor drugs development. (2) A novel accelerated gradient descent algorithm based MTL model was incorporated into the multitarget QSAR modeling with more efficiency in both convergence speed and learning accuracy. In summary, the computational pipeline presented here provided an efficient way to identify and design inhibitors that simultaneously and selectively bind to multiple targets multiple viruses with high affinity.
Future researches on multitarget QSAR analysis could be done to address the compound description issue with more kinds of feature descriptors. Also the investigations on the integration and fusion mechanisms of multiview feature spaces in compound representation could be conducted. Recently developed transfer learning technologies [36] in machine learning community may help to handle such cases efficiently. Furthermore, the underline mechanisms of HIVHCV coinfection as well as the synthesis of the coinhibitors based on our study are definitely worthy for longterm perusing.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
QL and HZ carried out the designing of the whole computational algorithm and drafted the manuscript. LL was responsible for the algorithm implementation. XC was responsible for the algorithm design. RZ and ZC conceived the study and participated in the design and coordination of the analyses. All authors read and approved the final manuscript.
Acknowledgements
The authors would like to thank Dr. Qiang Yang at Hong Kong University of Science and Technology for their suggestions. This work was supported in part by Project White Magnolia Funding, Shanghai (Grant No. 2010B127), Tongji Excellent Young Scientist Funding (Grant No. 2000219052), National Natural Science Foundation of China (Grant No. 30976611) and Research Fund for the Doctoral Program of Higher Education of China (Grant No.20100072120050)
References

Jorissen RN, Reddy G, Ali A, Altman MD, Chellappan S, Anjum SG, Tidor B, Schiffer CA, Rana TM, Gilson MK: Additivity in the analysis and design of HIV protease inhibitors.
J Med Chem 2009, 52(3):737754. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Johnston MI, Hoth DF: Present status and future prospects for HIV therapies.
Science 1993, 260(5112):1286. PubMed Abstract  Publisher Full Text

Rockstroh JK, Spengler U: HIV and hepatitis C virus coinfection.
The Lancet Infectious Diseases 2004, 4(7):437444. PubMed Abstract  Publisher Full Text

Solov'ev VP, Varnek A: AntiHIV Activity of HEPT, TIBO, and Cyclic Urea Derivatives: Structure Property Studies, Focused Combinatorial Library Generation, and Hits Selection Using Substructural Molecular Fragments Method.
J Chem Inf Comput Sci 2003, 43(5):17031719. PubMed Abstract  Publisher Full Text

Iyer M, Hopfinger AJ: Treating Chemical Diversity in QSAR Analysis: Modeling Diverse HIV1 Integrase Inhibitors Using 4D Fingerprints.
J Chem Inf Model 2007, 47(5):19451960. PubMed Abstract  Publisher Full Text

Patel PD, Patel MR, KaushikBasu N, Talele TT: 3D QSAR and molecular docking studies of benzimidazole derivatives as hepatitis C virus NS5B polymerase inhibitors.
J Chem Inf Model 2008, 48(1):4255. PubMed Abstract  Publisher Full Text

Nittoli T, Curran K, Insaf S, DiGrandi M, Orlowski M, Chopra R, Agarwal A, Howe AYM, Prashad A, Floyd MB: Identification of anthranilic acid derivatives as a novel class of allosteric inhibitors of hepatitis C NS5B polymerase.
J Med Chem 2007, 50(9):21082116. PubMed Abstract  Publisher Full Text

Harper S, Avolio S, Pacini B, Di Filippo M, Altamura S, Tomei L, Paonessa G, Di Marco S, Carfi A, Giuliano C: Potent inhibitors of subgenomic hepatitis C virus RNA replication through optimization of indoleNacetamide allosteric inhibitors of the viral NS5B polymerase.
J Med Chem 2005, 48(14):45474557. PubMed Abstract  Publisher Full Text

Chen KX, Nair L, Vibulbhan B, Yang W, Arasappan A, Bogen SL, Venkatraman S, Bennett F, Pan W, Blackman ML: SecondGeneration Highly Potent and Selective Inhibitors of the Hepatitis C Virus NS3 Serine Protease.
J Med Chem 2009, 52(5):13701379. PubMed Abstract  Publisher Full Text

Arasappan A, Padilla AI, Jao E, Bennett F, Bogen SL, Chen KX, Pike RE, Sannigrahi M, Soares J, Venkatraman S: Toward second generation hepatitis C virus NS3 serine protease inhibitors: discovery of novel P4 modified analogues with improved potency and pharmacokinetic profile.
Journal of medicinal chemistry 2009, 52(9):28062817. PubMed Abstract  Publisher Full Text

Bogen SL, Pan W, Ruan S, Nair LG, Arasappan A, Bennett F, Chen KX, Jao E, Venkatraman S, Vibulbhan B: Toward the BackUp of Boceprevir (SCH 503034): Discovery of New Extended P4Capped Ketoamide Inhibitors of Hepatitis C Virus NS3 Serine Protease with Improved Potency and Pharmacokinetic Profiles.
Journal of medicinal chemistry 2009, 52(12):36793688. PubMed Abstract  Publisher Full Text

Jenwitheesuk E, Horst JA, Rivas KL, Van Voorhis WC, Samudrala R: Novel paradigms for drug discovery: computational multitarget screening.
Trends in pharmacological sciences 2008, 29(2):6271. PubMed Abstract  Publisher Full Text

Pauwels R: New nonnucleoside reverse transcriptase inhibitors (NNRTIs) in development for the treatment of HIV infections.
Current Opinion in Pharmacology 2004, 4(5):437446. PubMed Abstract  Publisher Full Text

Reddy G, Ali A, Nalam MNL, Anjum SG, Cao H, Nathans RS, Schiffer CA, Rana TM: Design and Synthesis of HIV1 Protease Inhibitors Incorporating Oxazolidinones as P2/P2 'Ligands in Pseudosymmetric Dipeptide Isosteres.
J Med Chem 2007, 50(18):43164328. PubMed Abstract  Publisher Full Text

PrabuJeyabalan M, Nalivaika E, Schiffer CA: Substrate Shape Determines Specificity of Recognition for HIV1 Protease:: Analysis of Crystal Structures of Six Substrate Complexes.
Structure 2002, 10(3):369381. PubMed Abstract  Publisher Full Text

Chellappan S, Kairys V, Fernandes MX, Schiffer C, Gilson MK: Evaluation of the substrate envelope hypothesis for inhibitors of HIV1 protease.
Proteins: Structure, Function, and Bioinformatics 2007, 68(2):561567. Publisher Full Text

Schultz AK, Zhang M, Leitner T, Kuiken C, Korber B, Morgenstern B, Stanke M: A jumping profile Hidden Markov Model and applications to recombination sites in HIV and HCV genomes.
BMC bioinformatics 2006, 7(1):265. PubMed Abstract  BioMed Central Full Text  PubMed Central Full Text

Tan SL, Pause A, Shi Y, Sonenberg N: Hepatitis C therapeutics: current status and emerging strategies.
Nature Reviews Drug Discovery 2002, 1(11):867881. PubMed Abstract  Publisher Full Text

Franciscus A: Hepatitis C treatments in current clinical development.

Leach AR: Molecular modelling: principles and applications. AddisonWesley Longman Ltd; 2001.

Tomasselli AG, Heinrikson RL: Targeting the HIVprotease in AIDS therapy: a current clinical perspective1.
Biochimica et Biophysica Acta (BBA)Protein Structure and Molecular Enzymology 2000, 1477(12):189214. Publisher Full Text

Seden K, Back D, Khoo S: New directly acting antivirals for hepatitis C: potential for interaction with antiretrovirals.

Chen X, Pan W, Kwok JT, Carbonell JG: Accelerated gradient method for multitask sparse learning problem.

Labute P: A widely applicable set of descriptors.
Journal of Molecular Graphics and Modelling 2000, 18(45):464477. PubMed Abstract  Publisher Full Text

Xu J, Stevenson J: Druglike index: a new approach to measure druglike compounds and their diversity.
J Chem Inf Comput Sci 2000, 40(5):11771187. PubMed Abstract

Vilar S, Cozza G, Moro S: Medicinal chemistry and the molecular operating environment (MOE): application of QSAR and molecular docking to drug discovery.
Current Topics in Medicinal Chemistry 2008, 8(18):15551572. PubMed Abstract  Publisher Full Text

Cortes A, Cascante M, Cardenas ML, CornishBowden A: Relationships between inhibition constants, inhibitor concentrations for 50% inhibition and types of inhibition: new ways of analysing data.
Biochemical Journal 2001, 357(Pt 1):263. PubMed Abstract  Publisher Full Text  PubMed Central Full Text

Wu G, Yuan Y, Hodge CN: Determining appropriate substrate conversion for enzymatic assays in highthroughput screening.
Journal of Biomolecular Screening 2003, 8(6):694. PubMed Abstract  Publisher Full Text

Liu Q, Xu Q, Zheng VW, Xue H, Cao Z, Yang Q: Multitask learning for crossplatform siRNA efficacy prediction: an insilico study.

Liu Q, Che D, Huang Q, Cao Z, Zhu R: Multitarget QSAR Study in the Analysis and Design of HIV1 Inhibitors.

Selassie C, Verma RP: History of quantitative structureactivity relationships. In Burger's Medicinal Chemistry and Drug Discovery. Volume 1. sixth edition. Drug Discovery, John Wiley&Sons, Inc; 2003. PubMed Abstract

Duchi J, ShalevShwartz S, Singer Y, Chandra T: Efficient projections onto the l 1ball for learning in high dimensions.

Weaver S, Gleeson MP: The importance of the domain of applicability in QSAR modeling.
J Mol Graph Model 2008, 26(8):13151326. PubMed Abstract  Publisher Full Text

Tropsha A, Gramatica P, Gombar VK: The Importance of Being Earnest: Validation is the Absolute Essential for Successful Application and Interpretation of QSPR Models.
QSAR & Combinatorial Science 2003, 22(1):6977. PubMed Abstract  Publisher Full Text

Darnag R, Schmitzer A, Belmiloud Y, Villemin D, Jarid A, Chait A, Seyagh M, Cherqaoui D: QSAR Studies of HEPT Derivatives Using Support Vector Machines.
QSAR & Combinatorial Science 2009, 28(67):709718. PubMed Abstract  Publisher Full Text