SPADE Algorithm Research Articles

Confidence Analysis of Hotspot as Peat Forest Fire Indicator

Indonesia has recorded 14.9 million hectares of peat forest that continue to be deforested due to fire across Sumatra and Kalimantan. To operate a successful firefight, fast detection is a key element. Hotspot that appeared consecutively in more than two days is a strong indicator of fire existence. As the interest in data mining arose, an advanced technique can be implemented toward hotspot dataset into finding solutions. Many previous works have been carried out to mine sequence patterns and succeeded in determining as well as predicting areas with high occurrence of fire. However, none of the studies analyses the outliers, such as several hotspots which confidence decrease significantly in an adjacent interval of time. Confidence determines the quality of hotspot, with a value above 70% strongly indicates that fire spot exist. This study generated sequence patterns using the SPADE algorithm and analyses 21 hotspots considered as outliers using the Landsat-8 image. The result shows that 85.71% of hotspots have decreased confidence due to haze cover.

Identification of temporal condition patterns associated with pediatric obesity incidence using sequence mining and big data

BackgroundElectronic health records (EHRs) are potentially important components in addressing pediatric obesity in clinical settings and at the population level. This work aims to identify temporal condition patterns surrounding obesity incidence in a large pediatric population that may inform clinical care and childhood obesity policy and prevention efforts.MethodsEHR data from healthcare visits with an initial record of obesity incidence (index visit) from 2009 through 2016 at the Children’s Hospital of Philadelphia, and visits immediately before (pre-index) and after (post-index), were compared with a matched control population of patients with a healthy weight to characterize the prevalence of common diagnoses and condition trajectories. The study population consisted of 49,694 patients with pediatric obesity and their corresponding matched controls. The SPADE algorithm was used to identify common temporal condition patterns in the case population. McNemar’s test was used to assess the statistical significance of pattern prevalence differences between the case and control populations.ResultsSPADE identified 163 condition patterns that were present in at least 1% of cases; 80 were significantly more common among cases and 45 were significantly more common among controls (p < 0.05). Asthma and allergic rhinitis were strongly associated with childhood obesity incidence, particularly during the pre-index and index visits. Seven conditions were commonly diagnosed for cases exclusively during pre-index visits, including ear, nose, and throat disorders and gastroenteritis.ConclusionsThe novel application of SPADE on a large retrospective dataset revealed temporally dependent condition associations with obesity incidence. Allergic rhinitis and asthma had a particularly high prevalence during pre-index visits. These conditions, along with those exclusively observed during pre-index visits, may represent signals of future obesity. While causation cannot be inferred from these associations, the temporal condition patterns identified here represent hypotheses that can be investigated to determine causal relationships in future obesity research.

Effective Pattern Recognition Using Modified Prefix Span Algorithm

The advancement in the health and biotechnology has been produces huge amount of data that make records complex and expansive. In this paper data mining technique based on prefix span is used for diabetes disease detection and compared with the existing spade algorithm. The proposed algorithm uses MPV algorithm that discovers more patterns to detect disease accurately. The results will help in predicting quicker and more accurate disease so that it lead timely treatment of the patients.

PENENTUAN POLA SEKUENSIAL DATA TRANSAKSI PENJUALAN MENGGUNAKAN ALGORITMA SEQUENTIAL PATTREN DISCOVERY USING EQUIVALENT CLASSES (SPADE)

Transaction data is customer or customer data at a commercial or non-commercial institution that contains the consumer id, transaction time, and transaction items. From transaction data such as supermarket transactions, sequential patterns can be found to determine the interrelationship between items or items. Data if further processed or analyzed will produce information or knowledge that is important and valuable as a support in decision making. This study aims to determine the consumption patterns owned by customers and provide information that can be used in determining the layout of new store shelves. The SPADE algorithm is an algorithm for finding sequential patterns to break down the main problem into sub-problems that can be solved separately. Based on the result obtained it can be concluded that the application of the SPADE algorithm has the highest minimum support value that can still form maximal frequent sequences is 29%. The highest minimum support value of the SPADE algorithm is 0.2% with a maximum minimum confidence value of 81% and the number of rules formed is 1,118 Rule, but confidence is taken 60% up so that there are only 15 Rule. Whereas the Apriori algorithm has the highest minimum support value that can still form the maximum frequent sequences is 25%. The highest minimum support value of the apriori algorithm which can still form a rule is 0.3% with a maximum value of 88% minimum confidence and the number of rules formed as many as 494 Rule, but confidence is taken 60% up so that there are only 29 Rule.

Characterization of Leukocytes From HIV-ART Patients Using Combined Cytometric Profiles of 72 Cell Markers.

Motivation: Mass cytometry is a technique used to measure the intensity levels of proteins expressed by cells, at a single cell resolution. This technique is essential to characterize the phenotypes and functions of immune cell populations, but is currently limited to the measurement of 40 cell markers that restricts the characterization of complex diseases. However, algorithms and multi-tube cytometry techniques have been designed for combining phenotypic information obtained from different cytometric panels. The characterization of chronic HIV infection represents a good study case for multi-tube mass cytometry as this disease triggers a complex interactions network of more than 70 cell markers.Method: We collected whole blood from non-viremic HIV-infected patients on combined antiretroviral therapies and healthy donors. Leukocytes from each individual were stained using three different mass cytometry panels, which consisted of 35, 32, and 33 cell markers. For each patient and using the CytoBackBone algorithm, we combined phenotypic information from three different antibody panels into a single cytometric profile, reaching a phenotypic resolution of 72 markers. These high-resolution cytometric profiles were analyzed using SPADE and viSNE algorithms to decipher the immune response to HIV.Results: We detected an upregulation of several proteins in HIV-infected patients relative to healthy donors using our profiling of 72 cell markers. Among them, CD11a and CD11b were upregulated in PMNs, monocytes, mDCs, NK cells, and T cells. CD11b was also upregulated on pDCs. Other upregulated proteins included: CD38 on PMNs, monocytes, NK cells, basophils, B cells, and T cells; CD83 on monocytes, mDCs, B cells, and T cells; and TLR2, CD32, and CD64 on PMNs and monocytes. These results were validated using a mass cytometry panel of 25 cells markers.Impacts: We demonstrate here that multi-tube cytometry can be applied to mass cytometry for exploring, at an unprecedented level of details, cell populations impacted by complex diseases. We showed that the monocyte and PMN populations were strongly affected by the HIV infection, as CD11a, CD11b, CD32, CD38, CD64, CD83, CD86, and TLR2 were upregulated in these populations. Overall, these results demonstrate that HIV induced a specific environment that similarly affected multiple immune cells.

Mining Sequential Risk Patterns for Early Assessment of COPD

Nowadays, chronic diseases have been among the major concerns in medical fields since they may cause heavy burden on healthcare resources and disturb the quality of life. Chronic Obstructive Pulmonary Disease (COPD) is one kind of popular chronic diseases. COPD takes long period to evolve from mild symptoms (Stage I) to severe illness (Stage IV) and death. Earlier the disease is detected, the better the scope for effective treatment and improved control of symptom development. Therefore, the early of COPD is beneficial for better treatment. The paper examines the novel system for early appraisal on chronic illnesses by mining sequential risk patterns with interim data from diagnostic clinical records utilizing sequential rules mining and classification modelling systems. The system consists of four phases namely data pre-processing, risk pattern mining, classification modelling. SPADE algorithm and CBS algorithm used for risk pattern mining and classification. Decision tree algorithm is compared with the SPADE algorithm, and SPADE showing a better accuracy when comparing with decision tree.

Phenotipic Heterogeneity Studied By Spade Algorithm in Relapsed/Refractory Acute Myeloid Leukemia

Background.Acute myeloid leukemia (AML) is a heterogeneous disease with respect to cell phenotypes and clinical outcomes.Recently, detailed genetic studies have highlighted the mutational spectrum of clonal variability and increased our knowledge of biologic features of this disease. Nonetheless, characterization of AML cell heterogeneity by traditional Flow Cytometry (FC) is still ineffective because the hierarchical two-dimensional gating analysis of FC data is unable to distinguish all different sub-populations, as defined by the combination of multiple cell parameters. Here, we applied the SPADE algorithm (Ref. Qiu P., Nat Biotechnol, 2011), a new computational approach for dimensional reduction of FC data, in order to represent the intratumoral heterogeneity of AML at single-cell resolution in leukemias at the diagnosis vs. relapsed/refractory AMLs. The purpose of our study was to investigate the clonal dissimilarities of cell subsets as identified by SPADE analysis between diagnosis and relapse/refractory disease status in AML and to define the cell markers mainly associated to the clonal variation of tumor before and after standard therapies.Methods. 30AML patients were investigated at diagnosis and at relapse/ refractory (R/R) disease status by FC analysis.The SPADE algorithm was applied using Cytosplore software (Ref. Hollt. T, 2016) in order to depict the immunophenotype of clonal structures in AML before and after treatment. Each subpopulation was represented by colored circles (nodes) and described the substantial differences between and within individual leukemia subsets. At least a threshold of 7% of events was used to define significant clusters (node) and up to 50 target nodes were used to define the complexity of cell population. All nodes were hierarchically separated by relatedness into a “minimal spanning tree” in relation to the expression levels of 8 Ab markers used for this experiment in the FC panel. The readouts of spanning-tree progression analysis of normalized events (SPADE) reflected the cell heterogeneity underlying the fluorescence intensity of markers used in the FC analysis and defined the relationships among hierarchical cell subsets of leukemia before and after treatment. T-tests and Pearson's coefficient were used to compare antigenic distribution between diagnosis and R/R status. Results.Based on CD45 gating strategy, leukemic population of AML patients at diagnosis showed a median number of nodes significantly lower than that found at R/R disease status (n=4 vs. n= 6 nodes, p=0.0129)(Figure). These data defined an increase of different subpopulations in the R/R subset. Following antigens resulted mainly espressed on the entire study population: CD117, CD34, CD33, CD13, HLADR. The distribution of these markers were further investigated in order to define which cell parameters were able to discriminate the different subpopulations by CD45 expression. Indeed, a comparison of the number of nodes with highest antigenic expression (>90% of intensity value) between AML samples at diagnosis and R/R status was showed. CD117 and CD34 distribution showed a significant difference (p=0.0041 vs p=0.0139, respectively)(Figure) while no statistically significant values were observed among the others. However, only CD34+ nodes showed a significant linear growth between diagnosis and R/R status (p=0.0227).Conclusions. SPADE analysis allowed us to find an increase of subpopulations in R/R AML when compared to samples at diagnosis. CD34 and CD117 antigens mainly characterized these subpopulations, thus defining the cell markers mainly associated to the clonal variation. CD34 antigen was the widest distributed antigen among the new subpopulations, denoting a great instability of more immature clones. [Display omitted] DisclosuresNo relevant conflicts of interest to declare.

Toward deterministic and semiautomated SPADE analysis.

SPADE stands for spanning-tree progression analysis for density-normalized events. It combines downsampling, clustering and a minimum-spanning tree to provide an intuitive visualization of high-dimensional single-cell data, which assists with the interpretation of the cellular heterogeneity underlying the data. SPADE has been widely used for analysis of high-content flow cytometry data and CyTOF data. The downsampling and clustering components of SPADE are both stochastic, which lead to stochasticity in the tree visualization it generates. Running SPADE twice on the same data may generate two different tree structures. Although they typically lead to the same biological interpretation of subpopulations present in the data, robustness of the algorithm can be improved. Another avenue of improvement is the interpretation of the SPADE tree, which involves visual inspection of multiple colored versions of the tree based on expression of measured markers. This is essentially manual gating on the SPADE tree and can benefit from automated algorithms. This article presents improvements of SPADE in both aspects above, leading to a deterministic SPADE algorithm and a software implementation for semiautomated interpretation. © 2017 International Society for Advancement of Cytometry.

Hierarchical Pattern Mining with the Automata Processor

Mining complex patterns with hierarchical structures becomes more and more important to understand the underlying information in large and unstructured databases. When compared with a set-mining problem or a string-mining problem, the computation complexity to recognize a pattern with hierarchical structure, and the large associated search space, make hierarchical pattern mining (HPM) extremely expensive on conventional processor architectures. We propose a flexible, hardware-accelerated framework for mining hierarchical patterns with Apriori-based algorithms, which leads to multi-pass pruning strategies but exposes massive parallelism. Under this framework, we implemented two widely used HPM techniques, sequential pattern mining (SPM) and disjunctive rule mining (DRM) on the Automata Processor (AP), a hardware implementation of non-deterministic finite automata (NFAs). Two automaton-design strategies for matching and counting different types of hierarchical patterns, called linear design and reduction design, are proposed in this paper. To generalize automaton structure for SPM, the linear design strategy is proposed by flattening sequential patterns to plain strings to produce automaton design space and to minimize the overhead of reconfiguration. Up to 90\(\times \) and 29\(\times \) speedups are achieved by the AP-accelerated algorithm on six real-world datasets, when compared with the optimized multicore CPU and GPU GSP implementations, respectively. The proposed CPU-AP solution also outperforms the state-of-the-art PrefixSpan and SPADE algorithms on a multicore CPU by up to 452\(\times \) and 49\(\times \) speedups. The AP advantage grows further with larger datasets. For DRM, the reduction design strategy is adopted by applying reduction operation of AND, with on-chip Boolean units, on several parallel sub-structures for recognizing disjunctive items. This strategy allows implicit OR reduction on alternative items within a disjunctive item by utilizing bit-wise parallelism feature of the on-chip state units. The experiments show up to 614\(\times \) speedups of the proposed CPU-AP DRM solution over a sequential CPU algorithm on two real-world datasets. The experiments also show significant increase of CPU matching-and-counting time when increasing d-rule size or the number of alternative items. However, in practical cases, the AP solution runs hundreds of times faster in matching and counting than the CPU solution, and keeps constant processing time despite the increasing complexity of disjunctive rules.

SPADEVizR: an R package for visualization, analysis and integration of SPADE results.

MotivationFlow, hyperspectral and mass cytometry are experimental techniques measuring cell marker expressions at the single cell level. The recent increase of the number of markers simultaneously measurable has led to the development of new automatic gating algorithms. Especially, the SPADE algorithm has been proposed as a novel way to identify clusters of cells having similar phenotypes in high-dimensional cytometry data. While SPADE or other cell clustering algorithms are powerful approaches, complementary analysis features are needed to better characterize the identified cell clusters.ResultsWe have developed SPADEVizR, an R package designed for the visualization, analysis and integration of cell clustering results. The available statistical methods allow highlighting cell clusters with relevant biological behaviors or integrating them with additional biological variables. Moreover, several visualization methods are available to better characterize the cell clusters, such as volcano plots, streamgraphs, parallel coordinates, heatmaps, or distograms. SPADEVizR can also generate linear, Cox or random forest models to predict biological outcomes, based on the cell cluster abundances. Additionally, SPADEVizR has several features allowing to quantify and to visualize the quality of the cell clustering results. These analysis features are essential to better interpret the behaviors and phenotypes of the identified cell clusters. Importantly, SPADEVizR can handle clustering results from other algorithms than SPADE.Availability and ImplementationSPADEVizR is distributed under the GPL-3 license and is available at https://github.com/tchitchek-lab/SPADEVizR.Supplementary information Supplementary data are available at Bioinformatics online.

FP-Split SPADE-An Algorithm for Finding Sequential Patterns

Sequential Pattern Mining (SPM) is one of the key areas in Web Usage Mining (WUM) with broad applications such as analyzing customer behavior from weblog files. The current algorithms in this area can be classified into two broad areas, namely, apriori-based and pattern-growth based. Apriori based algorithms for mining sequential patterns need to scan the database many times as they focus on candidate generation and test approach. A lot of research has been done so far, but even the best apriori based algorithm for SPM in terms of number of database scans is SPADE that scans the database three times for discovering sequential patterns. Pattern growth based algorithms avoid the candidate generation step and the best pattern growth algorithm known so far is Prefix Span that needs to scan the database at least twice. In this paper, a novel algorithm for SPM is proposed called FP-Split SPADE that reduced the database scan to only one by creating an FP-Split tree and applying SPADE algorithm on the tree instead on sequence database that greatly improved the efficiency of mining sequential patterns.

SPADEVizR: an R package for visualization, analysis and integration of SPADE results

Motivation Flow, hyperspectral and mass cytometry are experimental techniques measuring cell marker expressions at the single cell level. The recent increase of the number of markers simultaneously measurable has led to the development of new automatic gating algorithms. Especially, the SPADE algorithm has been proposed as a novel way to identify clusters of cells having similar phenotypes in high-dimensional cytometry data. While SPADE or other cell clustering algorithms are powerful approaches, complementary analysis features are needed to better characterize the identified cell clusters. Results We have developed SPADEVizR, an R package designed for the visualization, analysis and integration of cell clustering results. The available statistical methods allow highlighting cell clusters with relevant biological behaviors or integrating them with additional biological variables. Moreover, several visualization methods are available to better characterize the cell clusters, such as volcano plots, streamgraphs, parallel coordinates, heatmaps, or distograms. SPADEVizR can also generate linear, Cox or random forest models to predict biological outcomes, based on the cell cluster abundances. Additionally, SPADEVizR has several features allowing to quantify and to visualize the quality of the cell clustering results. These analysis features are essential to better interpret the behaviors and phenotypes of the identified cell clusters. Importantly, SPADEVizR can handle clustering results from other algorithms than SPADE. Availability and implementation SPADEVizR is distributed under the GPL-3 license and is available at https://github.com/tchitchek-lab/SPADEVizR . Contact nicolas.tchitchek@gmail.com. Supplementary information Supplementary data are available at Bioinformatics online.

Sequence Mining of Comorbid Neurodevelopmental Disorders Using the SPADE Algorithm.

Understanding the progression of comorbid neurodevelopmental disorders (NDD) during different critical time periods may contribute to our comprehension of the underlying pathophysiology of NDDs. The objective of our study was to identify frequent temporal sequences of developmental diagnoses in noisy patient data. We used a data set of 2810 patients, documenting NDD diagnoses given to them by an NDD expert at a child developmental center during multiple visits at different ages. Extensive preprocessing steps were developed in order to allow the data set to be processed by an efficient sequence mining algorithm (SPADE). The discovered sequences were validated by cross validation for 10 iterations; all correlation coefficients for support, confidence and lift measures were above 0.75 and their proportions were similar. No signifi- cant differences between the distributions of sequences were found using Kolmogorov-Smirnov test. We have demonstrated the feasibility of using the SPADE algorithm for discovery of valid temporal sequences of comorbid disorders in children with NDDs. The identification of such sequences would be beneficial from clinical and research perspectives. Moreover, these sequences could serve as features for developing a full-fledged temporal predictive model.

Activation of the Erythroid Transcriptional Program in Murine Adult Bone Marrow Takes Place during a Faster, Shorter S Phase and Is Dependent on S Phase Progression

We previously characterized an S phase-dependent commitment switch in mouse fetal liver. Specifically, the upregulation of CD71 in the transition from Subset 0 (S0, Lin-CD71medium) to Subset 1 (S1, Lin-CD71high) identifies cells in early S phase of the last colony-forming-unit-erythroid (CFUe) generation, undergoing key differentiation events, including onset of Erythropoietin (Epo) dependence, GATA-1 activation and reconfiguration of chromatin at the §-globin locus control region. This commitment switch requires S phase progression, and is regulated by the S-phase- dependent downregulation of the transcription factor PU.1, a GATA-1 antagonist. Of note, the specific cell cycle S phase in which this commitment switch takes place differs from S phase in previous cycles, in that it is shorter, with a 50% faster rate of nucleotide incorporation into DNA.Here, we investigate whether a similar commitment switch takes place in adult bone-marrow. Using fresh bone-marrow from mice in the basal state or following Epo injection, we determined colony-forming potential, transcriptional profiles and cell cycle status of Kit+Lin- subsets defined by cell surface markers that have been previously implicated as enriched in the megakaryocytic-erythrocytic lineages: CD105, CD150, CD55, CD71, in conjunction with a PU.1-GFP reporter mouse.Using quantitative RT-PCR together with flow-cytometric analysis and the SPADE algorithm we identified a linear erythroid developmental sequence of Epo-responsive Kit+Lin-CD55+ bone-marrow subsets, which gradually declined in PU.1, transiently increased CD150, and upregulated both CD105 and CD71. The loss of PU.1 coincided with increased GATA1 levels and the transcriptional activation of erythroid genes. CFUe activity peaked within Kit+Lin-CD55+ cells that were also CD105+, where colonies of other lineage potential were rare (Fig 1).Upregulation of CD71 in erythroid cells indicates the onset of Epo receptor signaling and Epo dependence. We found that in Kit+Lin-CD55+ bone-marrow in vivo, high levels of CD71 were highly correlated with S phase (Fig 2). Furthermore, the rate of incorporation of the nucleotide analogue BrdU was substantially higher in CD71 -high cells, suggesting a rapid, short S phase. To functionally examine the role of S phase, we isolated Kit+Lin-CD55+CD105+ cells that were CD71 low, and followed their differentiation in vitro. These cells upregulate CD71 within 10 to 16 hours, followed by upregulation of Ter119. Upregulation of CD71 was Epo dependent, and took place in S phase; further, it was reversibly inhibited by the DNA polymerase inhibitor, Aphidicolin, suggesting dependence on S phase progression.We conclude that adult bone marrow Kit+Lin-CD55+CD105+ are at the CFUe developmental stage, and undergo an Epo and S-phase dependent commitment switch that activates the erythroid transcriptional and differentiation program. S phase during this commitment switch is characterized by a fast DNA synthesis rate. [Display omitted] [Display omitted] DisclosuresNo relevant conflicts of interest to declare.

Unsupervised Discovery of Early Markers of Erythroid Maturation in Human Donor Marrow

Antigenic expression during erythroid maturation is not well understood. Despite the large number of surface antigens currently described for leukocytes, only a few (e.g. CD36, CD45, CD71, CD117, and CD235a) have been well characterized on erythroid cells. Here, we apply novel bioinformatic tools to select antigens with high potential for identifying successive stages of erythroid maturation in an unsupervised and unbiased manner from a high dimensional dataset. In brief, flow cytometry was performed on 3 normal donor bone marrows using Becton-Dickinson lyoplates containing a total of 275 unique antibodies and 8 gating reagents (CD15, CD19, CD34, CD38, CD45, CD71, CD117, and CD123). Each of the 275 independent data files from the same donor was aligned using the gating reagents and a weighted nearest neighbor algorithm in order to synthesize a flow data set with 283 antigens and 50,000 events. We applied a modified SPADE algorithm to generate a maturational path for cells of the erythroid lineage starting with CD34+/CD38- progenitors through the mature erythrocyte stage. A non-parametric Kruskal–Wallis test was used to identify and rank antigens that show differential expression along the erythroid maturational sequence.To discern antigens having the greatest discrimination for early vs. late erythroid maturation, we identified 15 antigens common to the top 25 most differentially expressed antigens from each donor sample. As expected, our method correctly identified the three erythroid gating reagents (CD45, CD71 and CD117) as well as previously a previously described erythroid associated antigen (CD36). A few of the identified antigens have been described are less common but also known to be differentially expressed on erythroid cells (e.g. integrin members CD29, CD44, and CD49d), however several additional novel antigens were also identified with strong differential expression including CD46, CD58, CD81, CD98, CD99, CD164, CD220 and CD321 (Figure 1). Of note, CD49d, CD98 and CD164 are of particular interest as they appear to be gradually lost during maturation from the early normoblast through the mature erythrocyte stages. This is in contrast with the other antigens that show high expression on pronormoblasts—with rapid decline during the early normoblast stage to the level of erythrocytes. We examined selected antigens from the synthetic data of all 3 samples using manual gating, and interestingly CD99 retained expression among the erythroid cells to late stage normoblasts in 2 of the 3 samples, suggesting some element of phenotypic variability among normal individuals for this antigen.Additionally, several antigens appear useful in distinguishing erythroid lineage cells from non-erythroid precursors including beta-2 microglobulin, CD50, and HLA-ABC which all showed ubiquitous expression in non-erythroid precursors with low to absent expression among cells of the erythroid lineage.In summary, using a novel unbiased method for large-scale antigen discovery, we have identified multiple novel antigens that are differentially down regulated with progressive erythroid maturation and appear useful in further delineating erythroid progenitor maturation. Additional work is underway to correlate these antigens with our morphologic understanding of erythroid maturation. Further work in characterizing these changes in myeloid stem cell disorders is on going and may be of diagnostic utility in the diagnosis of myelodysplastic syndromes. [Display omitted] DisclosuresNo relevant conflicts of interest to declare.

T cell responses to Epstein Barr Virus infections in pediatric organ transplant recipients (P2196)

Abstract Pediatric organ transplant recipients are at highest risk of developing morbidity and mortality from EBV infection. There is broad consensus that T cell immunity is important in controlling viral replication and expansion of infected B cells. In this study, we simultaneously analyze virus-specific T cells’ phenotypes, functions, and epitope specificity using 37-parameter mass cytometry in the setting of pediatric organ transplant recipients, followed over the first two years post-transplant. We compared T cell responses to peptide pools derived from lytic and latent EBV antigens in both healthy EBV+ donors and pediatric transplant patients. The results suggest that healthy EBV+ donors can mount CD8+ T cell responses to both lytic and latent EBV antigens, with production of multiple cytokines. In contrast, pediatric patients have much weaker T cell responses, especially to lytic antigens. Based on current data, we hypothesize a polyfunctional T cell response to lytic antigens may preferentially protect against EBV viremia. We also use the SPADE algorithm for unsupervised clustering to compare the distribution of different T cell subsets in cytokine-producing cells. In addition, we analyze these T cells’ functional capacity for cytokine production by visualization in a heatmap. By employing highly multi-parameter mass cytometry and longitudinal sampling, we have created a more detailed picture of the T cell responses to EBV in the setting of pediatric transplant recipients.

Mining Sequential Patterns in Dense Databases

Sequential pattern mining is an important data mining problem with broad applications, including the analysis of customer purchase patterns, Web access patterns, DNA analysis, and so on. We show on dense databases, a typical algorithm like Spade algorithm tends to lose its efficiency. Spade is based on the used of lists containing the localization of the occurrences of pattern in the sequences and these lists are not appropriated in the case of dense databases. In this paper we present an adaptation of the wellknown diffset data representation [12] with Spade algorithm. The new version is called dSpade. Since diffset shows high performance for mining frequent itemsets in dense transactional databases, experimental evaluation shows that dSpade is suitable for mining dense sequence databases.