Fabio Fumarola

Novel targeting of phospho-cMET overcomes drug resistance and induces antitumor activity in multiple myeloma

Purpose: The aim of the study was to verify the hypothesis that the cMet oncogene is implicated in chemio- and novel drug resistance in multiple myeloma. Experimental Design: We have evaluated the expression levels of cMET/phospho-cMET (p-cMET) and the activity of the novel selective p-cMET inhibitor (SU11274) in multiple myeloma cells, either sensitive (RPMI-8226 and MM.1S) or resistant (R5 and MM.1R) to anti–multiple myeloma drugs, in primary plasma cells and in multiple myeloma xenograft models. Results: We found that resistant R5 and MM.1R cells presented with higher cMET phosphorylation, thus leading to constitutive activation of cMET-dependent signaling pathways. R5 cells exhibited a higher susceptibility to the SU11274 inhibitory effects on viability, proliferation, chemotaxis, adhesion, and to its apoptogenic effects. SU11274 was able to revert drug resistance in R5 cells. R5 but not RPMI-8226 cells displayed cMET-dependent activation of mitogen-activated protein kinase pathway. The cMET and p-cMET expression was higher on plasma cells from patients with multiple myeloma at relapse or on drug resistance than on those from patients at diagnosis, complete/partial remission, or from patients with monoclonal gammopathy of unknown significance. Viability, chemotaxis, adhesion to fibronectin or paired bone marrow stromal cells of plasma cells from relapsed or resistant patients was markedly inhibited by SU11274. Importantly, SU11274 showed higher therapeutic activity in R5- than in RPMI-8226–induced plasmocytomas. In R5 tumors, it caused apoptosis and necrosis and reverted bortezomib resistance. Conclusion: Our findings suggest that the cMET pathway is constitutively activated in relapsed and resistant multiple myeloma where it may also be responsible for induction of drug resistance, thus providing the preclinical rationale for targeting cMET in patients with relapsed/refractory multiple myeloma. Clin Cancer Res; 19(16); 4371–82. ©2013 AACR.

Extracting general lists from web documents: a hybrid approach

The problem of extracting structured data (i.e. lists, record sets, tables, etc.) from the Web has been traditionally approached by taking into account either the underlying markup structure of a Web page or the visual structure of the Web page. However, empirical results show that considering the HTML structure and visual cues of a Web page independently do not generalize well. We propose a new hybrid method to extract general lists from the Web. It employs both general assumptions on the visual rendering of lists, and the structural representation of items contained in them. We show that our method significantly outperforms existing methods across a varied Web corpus.

Completion Time and Next Activity Prediction of Processes Using Sequential Pattern Mining

Process mining is a research discipline that aims to discover, monitor and improve real processing using event logs. In this paper we describe a novel approach that (i) identifies partial process models by exploiting sequential pattern mining and (ii) uses the additional information about the activities matching a partial process model to train nested prediction models from event logs. Models can be used to predict the next activity and completion time of a new (running) process instance. We compare our approach with a model based on Transition Systems implemented in the ProM5 Suite and show that the attributes in the event log can improve the accuracy of the model without decreasing performances. The experimental results show how our algorithm improves of a large margin ProM5 in predicting the completion time of a process, while it presents competitive results for next activity prediction.

Unexpected results in automatic list extraction on the web

The discovery and extraction of general lists on the Web continues to be an important problem facing theWeb mining community. There have been numerous studies that claim to automatically extract structured data (i.e. lists, record sets, tables, etc.) from the Web for various purposes. Our own recent experiences have shown that the list-finding methods used as part of these larger frameworks do not generalize well and therefore ought to be reevaluated. This paper briefly describes some of the current approaches, and tests them on various list-pages. Based on our findings, we conclude that analyzing aWeb pages DOM-structure is not sufficient for the general list finding task.

Predictive Modeling of PV Energy Production: How to Set Up the Learning Task for a Better Prediction?

In this paper, we tackle the problem of power prediction of several photovoltaic (PV) plants spread over an extended geographic area and connected to a power grid. The paper is intended to be a comprehensive study of one-day ahead forecast of PV energy production along several dimensions of analysis: 1) The consideration of the spatio-temporal autocorrelation, which characterizes geophysical phenomena, to obtain more accurate predictions. 2) The learning setting to be considered, i.e., using simple output prediction for each hour or structured output prediction for each day. 3) The learning algorithms: We compare artificial neural networks, most often used for PV prediction forecast, and regression trees for learning adaptive models. The results obtained on two PV power plant datasets show that: taking into account spatio/temporal autocorrelation is beneficial; the structured output prediction setting significantly outperforms the nonstructured output prediction setting; and regression trees provide better models than artificial neural networks.

HyLiEn: a hybrid approach to general list extraction on the web

We consider the problem of automatically extracting general lists from the web. Existing approaches are mostly dependent upon either the underlying HTML markup or the visual structure of the Web page. We present HyLiEn an unsupervised, Hybrid approach for automatic List discovery and Extraction on the Web. It employs general assumptions about the visual rendering of lists, and the structural representation of items contained in them. We show that our method significantly outperforms existing methods.

WINACS: construction and analysis of web-based computer science information networks

WINACS (Web-based Information Network Analysis for Computer Science) is a project that incorporates many recent, exciting developments in data sciences to construct a Web-based computer science information network and to discover, retrieve, rank, cluster, and analyze such an information network. With the rapid development of the Web, huge amounts of information are available in the form of Web documents, structures, and links. It has been a dream of the database and Web communities to harvest such information and reconcile the unstructured nature of the Web with the neat, semi-structured schemas of the database paradigm. Taking computer science as a dedicated domain, WINACS first discovers related Web entity structures, and then constructs a heterogeneous computer science information network in order to rank, cluster and analyze this network and support intelligent and analytical queries.

CloFAST: closed sequential pattern mining using sparse and vertical id-lists

Sequential pattern mining is a computationally challenging task since algorithms have to generate and/or test a combinatorially explosive number of intermediate subsequences. In order to reduce complexity, some researchers focus on the task of mining closed sequential patterns. This not only results in increased efficiency, but also provides a way to compact results, while preserving the same expressive power of patterns extracted by means of traditional (non-closed) sequential pattern mining algorithms. In this paper, we present CloFAST, a novel algorithm for mining closed frequent sequences of itemsets. It combines a new data representation of the dataset, based on sparse id-lists and vertical id-lists, whose theoretical properties are studied in order to fast count the support of sequential patterns, with a novel one-step technique both to check sequence closure and to prune the search space. Contrary to almost all the existing algorithms, which iteratively alternate itemset extension and sequence extension, CloFAST proceeds in two steps. Initially, all closed frequent itemsets are mined in order to obtain an initial set of sequences of size 1. Then, new sequences are generated by directly working on the sequences, without mining additional frequent itemsets. A thorough performance study with both real-world and artificially generated datasets empirically proves that CloFAST outperforms the state-of-the-art algorithms, both in time and memory consumption, especially when mining long closed sequences.

FAST sequence mining based on sparse id-lists

Sequential pattern mining is an important data mining task with applications in basket analysis, world wide web, medicine and telecommunication. This task is challenging because sequence databases are usually large with many and long sequences and the number of possible sequential patterns to mine can be exponential. We proposed a new sequential pattern mining algorithm called FAST which employs a representation of the dataset with indexed sparse id-lists to fast counting the support of sequential patterns. We also use a lexicographic tree to improve the efficiency of candidates generation. FAST mines the complete set of patterns by greatly reducing the effort for support counting and candidate sequences generation. Experimental results on artificial and real data show that our method outperforms existing methods in literature up to an order of magnitude or two for large datasets.

Growing parallel paths for entity-page discovery

In this paper, we use the structural and relational information on the Web to find entity-pages. Specifically, given a Web site and an entity-page (e.g., department and faculty member homepage) we seek to find all of the entity-pages of the same type (e.g., all faculty members in the department). To do this, we propose a web structure mining method which grows parallel paths through the web graph and DOM trees. We show that by utilizing these parallel paths we can efficiently discover all entity-pages of the same type. Finally, we demonstrate the accuracy of our method with a case study on various domains.