Liu, Hongfang

Abstract High-throughput data is increasingly being integrated into the fields of biology, molecular biology, and pharmacology. However, a difficult problem has been the rapid and fluid access to and integration of the data, which tends to reside in huge unwieldy databases. One set of cell lines with substantial potential for benefit from this type of access and integration is the NCI-60 cancerous cell lines. We present here a set of tools within our CellMiner web-application designed to address this need for the areas of transcript expression, microRNA expression, gene DNA copy number, variant status, and drug activity. CellMiner allows the user to rapidly access data for relative levels of transcript expression for 26,065 genes, 360 microRNAs, and 20,602 compounds including 102 Food and Drug Administration (FDA)-approved drugs. These levels in turn create patterns across the NCI-60 that can be compared to one another using our “pattern match” tool. Together, these tools allow one to query the data for potential relationships between these parameters, in a manner specific to a users area of expertise and interest, in a rapid and flexible manner without the need for expertise in computer science or bioinformatics. Comparisons of transcript/drug will be demonstrated with SLFN11/topotecan; variant/drug with BRAF V600E/vemurafinib; colon tissue-of origin specific genes with TRIM15, RNF43, and VIL1; and DNA copy number change with CDKN2A (p16). The data are publicly available at http://discover.nci.nih.gov/cellminer. Citation Format: William C. Reinhold, Margot Sunshine, Sudir Varma, Hongfang Liu, Ogan Abaan, Paul Meltzer, Joel Morris, Kurt Kohn, James Doroshow, Yves Pommier. Web-based access using CellMiner for gene expression, DNA copy number, microRNA transcript levels, variant status, drug activity, and their pattern comparisons for the NCI-60. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 3177. doi:10.1158/1538-7445.AM2013-3177

Because of privacy concerns and the expense involved in creating an annotated corpus, the existing small-annotated corpora might not have sufficient examples for learning to statistically extract all the named-entities precisely. In this work, we evaluate what value may lie in automatically generated features based on distributional semantics when using machine-learning named entity recognition (NER). The features we generated and experimented with include n-nearest words, support vector machine (SVM)-regions, and term clustering, all of which are considered distributional semantic features. The addition of the n-nearest words feature resulted in a greater increase in F-score than by using a manually constructed lexicon to a baseline system. Although the need for relatively small-annotated corpora for retraining is not obviated, lexicons empirically derived from unannotated text can not only supplement manually created lexicons, but also replace them. This phenomenon is observed in extracting concepts from both biomedical literature and clinical notes.

Abstract High-throughput data is increasingly being integrated into the fields of biology, molecular biology, and pharmacology. However, a difficult problem has been the rapid and fluid access to and integration of this data, which tends to reside in huge unwieldy databases. One set of cell lines with substantial potential for benefit from this type of access and integration is the NCI-60 cancerous cell lines. We present here a set of tools within our CellMiner web-application designed to address this need for the areas of transcript expression, microRNA expression, gene DNA copy number, and drug activity. CellMiner allows the user to rapidly access data for relative levels of transcript expression for 18,532 genes, 360 microRNAs, and 16,861 compounds including 91 Food and Drug Administration (FDA)-approved drugs. These levels in turn create patterns across the NCI-60 that can be compared to one another using our “pattern match” tool. Together, these tools allow one to query the data for potential relationships between these parameters, in a manner specific to a users area of expertise and interest, in a rapid and flexible manner without the need for expertise in computer science or bioinformatics. The output will be demonstrated with cancer interesting microRNAs 18a, 21, 29a, 30c-2, 30d, and 31, cancer interesting genes ABCB1, ATM, ATR, BRCA2, and TP53, and the clinically relevant drugs belinostat, crizotinib, pazopanib, selumetinib and tanespimycin. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 2991. doi:1538-7445.AM2012-2991

This paper describes the sentiment classification system developed by the Mayo Clinic team for the 2011 I2B2/VA/Cincinnati Natural Language Processing (NLP) Challenge. The sentiment classification task is to assign any pertinent emotion to each sentence in suicide notes. We have implemented three systems that have been trained on suicide notes provided by the I2B2 challenge organizer–-a machine learning system, a rule-based system, and a system consisting of a combination of both. Our machine learning system was trained on re-annotated data in which apparently inconsistent emotion assignment was adjusted. Then, the machine learning methods by RIPPER and multinomial Naïve Bayes classifiers, manual pattern matching rules, and the combination of the two systems were tested to determine the emotions within sentences. The combination of the machine learning and rule-based system performed best and produced a micro-average F-score of 0.5640.