Siddhartha Jain

Large neighborhood search for dial-a-ride problems

Dial-a-Ride problems (DARPs) arise in many urban transportation applications. The core of a DARP is a pick and delivery routing with multiple vehicles in which customers have ride-time constraints and routes have a maximum duration. This paper considers DARPs for which the objective is to minimize the routing cost, a complex optimization problem which has been studied extensively in the past. State-of-the-art approaches include sophisticated tabu search and variable neighborhood search. This paper presented a simple constraint-based large neighborhood search, which uses constraint programming repeatedly to find good reinsertions for randomly selected sets of customers. Experimental evidence shows that the approach is competitive in finding best-known solutions and reaches high-quality solutions significantly faster than the state of the art.

Using neural networks for reducing the dimensions of single-cell RNA-Seq data

Abstract While only recently developed, the ability to profile expression data in single cells (scRNA-Seq) has already led to several important studies and findings. However, this technology has also raised several new computational challenges. These include questions about the best methods for clustering scRNA-Seq data, how to identify unique group of cells in such experiments, and how to determine the state or function of specific cells based on their expression profile. To address these issues we develop and test a method based on neural networks (NN) for the analysis and retrieval of single cell RNA-Seq data. We tested various NN architectures, some of which incorporate prior biological knowledge, and used these to obtain a reduced dimension representation of the single cell expression data. We show that the NN method improves upon prior methods in both, the ability to correctly group cells in experiments not used in the training and the ability to correctly infer cell type or state by querying a database of tens of thousands of single cell profiles. Such database queries (which can be performed using our web server) will enable researchers to better characterize cells when analyzing heterogeneous scRNA-Seq samples.

A complete multi-valued SAT solver

We present a new complete multi-valued SAT solver, based on current state-of-the-art SAT technology. It features watched literal propagation and conflict driven clause learning. We combine this technology with state-of-the-art CP methods for branching and introduce quantitative supports which augment the watched literal scheme with a watched domain size scheme. Most importantly, we adapt SAT nogood learning for the multi-valued case and demonstrate that exploiting the knowledge that each variable must take exactly one out of many values can lead to much stronger nogoods. Experimental results assess the benefits of these contributions and show that solving multi-valued SAT directly often works better than reducing multi-valued constraint problems to SAT.

Temporal transcriptional response to latency reversing agents identifies specific factors regulating HIV-1 viral transcriptional switch

BackgroundLatent HIV-1 reservoirs are identified as one of the major challenges to achieve HIV-1 cure. Currently available strategies are associated with wide variability in outcomes both in patients and CD4+ T cell models. This underlines the critical need to develop innovative strategies to predict and recognize ways that could result in better reactivation and eventual elimination of latent HIV-1 reservoirs.Results and discussionIn this study, we combined genome wide transcriptome datasets post activation with Systems Biology approach (Signaling and Dynamic Regulatory Events Miner, SDREM analyses) to reconstruct a dynamic signaling and regulatory network involved in reactivation mediated by specific activators using a latent cell line. This approach identified several critical regulators for each treatment, which were confirmed in follow-up validation studies using small molecule inhibitors. Results indicate that signaling pathways involving JNK and related factors as predicted by SDREM are essential for virus reactivation by suberoylanilide hydroxamic acid. ERK1/2 and NF-κB pathways have the foremost role in reactivation with prostratin and TNF-α, respectively. JAK-STAT pathway has a central role in HIV-1 transcription. Additional evaluation, using other latent J-Lat cell clones and primary T cell model, also confirmed that many of the cellular factors associated with latency reversing agents are similar, though minor differences are identified. JAK-STAT and NF-κB related pathways are critical for reversal of HIV-1 latency in primary resting T cells.ConclusionThese results validate our combinatorial approach to predict the regulatory cellular factors and pathways responsible for HIV-1 reactivation in latent HIV-1 harboring cell line models. JAK-STAT have a role in reversal of latency in all the HIV-1 latency models tested, including primary CD4+ T cells, with additional cellular pathways such as NF-κB, JNK and ERK 1/2 that may have complementary role in reversal of HIV-1 latency.

Reconstructing the temporal progression of HIV-1 immune response pathways

Motivation: Most methods for reconstructing response networks from high throughput data generate static models which cannot distinguish between early and late response stages. Results: We present TimePath, a new method that integrates time series and static datasets to reconstruct dynamic models of host response to stimulus. TimePath uses an Integer Programming formulation to select a subset of pathways that, together, explain the observed dynamic responses. Applying TimePath to study human response to HIV-1 led to accurate reconstruction of several known regulatory and signaling pathways and to novel mechanistic insights. We experimentally validated several of TimePaths’ predictions highlighting the usefulness of temporal models. Availability and Implementation: Data, Supplementary text and the TimePath software are available from http://sb.cs.cmu.edu/timepath Contact: zivbj@cs.cmu.edu Supplementary information: Supplementary data are available at Bioinformatics online.

Transcriptome analyses identify key cellular factors associated with HIV-1 associated neuropathogenesis in infected men.

Objective: HIV-1 viral proteins and host inflammatory factors have a direct role in neuronal toxicity in vitro; however, the contribution of these factors in vivo in HIV-1-associated neurocognitive disorder (HAND) is not fully understood. We applied novel Systems Biology approaches to identify specific cellular and viral factors and their related pathways that are associated with different stages of HAND. Design: A cross-sectional study of individuals enrolled in the Multicenter AIDS Cohort Study including HIV-1-seronegative (N = 36) and HIV-1-seropositive individuals without neurocognitive symptoms (N = 16) or with mild neurocognitive disorder (MND) (N = 8) or HIV-associated dementia (HAD) (N = 16). Methods: A systematic evaluation of global transcriptome of peripheral blood mononuclear cells (PBMCs) obtained from HIV-1-seronegative individuals and from HIV-1-positive men without neurocognitive symptoms, or MND or HAD was performed. Results: MND and HAD were associated with specific changes in mRNA transcripts and microRNAs in PBMCs. Comparison of upstream regulators and TimePath analyses identified specific cellular factors associated with MND and HAD, whereas HIV-1 viral proteins played a greater role in HAD. In addition, expression of specific microRNAs – miR-let-7a, miR-124, miR-15a and others – were found to correlate with mRNA gene expression and may have a potential protective role in asymptomatic HIV-1-seropositive individuals by regulating cellular signal transduction pathways downstream of chemokines and cytokines. Conclusion: These results identify signature transcriptome changes in PBMCs associated with stages of HAND and shed light on the potential contribution of host cellular factors and viral proteins in HAND development.

Using Neural Networks To Improve Single-Cell RNA-Seq Data Analysis

While only recently developed, the ability to profile expression data in single cells (scRNA-Seq) has already led to several important studies and findings. However, this technology has also raised several new computational challenges including questions related to handling the noisy and sometimes incomplete data, how to identify unique group of cells in such experiments and how to determine the state or function of specific cells based on their expression profile. To address these issues we develop and test a method based on neural networks (NN) for the analysis and retrieval of single cell RNA-Seq data. We tested various NN architectures, some biologically motivated, and used these to obtain a reduced dimension representation of the single cell expression data. We show that the NN method improves upon prior methods in both, the ability to correctly group cells in experiments not used in the training and the ability to correctly infer cell type or state by querying a database of tens of thousands of single cell profiles. Such database queries (which can be performed using our web server) will enable researchers to better characterize cells when analyzing heterogeneous scRNA-Seq samples. Supporting website: http://sb.cs.cmu.edu/scnn/ Password for accessing the retrieval task webserver: scRNA-Seq

Upper bounds on the number of solutions of binary integer programs

We present a new method to compute upper bounds of the number of solutions of binary integer programming (BIP) problems. Given a BIP, we create a dynamic programming (DP) table for a redundant knapsack constraint which is obtained by surrogate relaxation. We then consider a Lagrangian relaxation of the original problem to obtain an initial weight bound on the knapsack. This bound is then refined through subgradient optimization. The latter provides a variety of Lagrange multipliers which allow us to filter infeasible edges in the DP table. The number of paths in the final table then provides an upper bound on the number of solutions. Numerical results show the effectiveness of our counting framework on automatic recording and market split problems.