A Comprehensive Investigation of Active Learning Strategies for Conducting Anti-Cancer Drug Screening.

It is well-known that cancers of the same histology type can respond differently to a treatment. Thus, computational drug response prediction is of paramount importance for both preclinical drug screening studies and clinical treatment design. To build drug response prediction models, treatment response data need to be generated through screening experiments and used as input to train the prediction models. In this study, we investigate various active learning strategies of selecting experiments to generate response data for the purposes of (1) improving the performance of drug response prediction models built on the data and (2) identifying effective treatments. Here, we focus on constructing drug-specific response prediction models for cancer cell lines. Various approaches have been designed and applied to select cell lines for screening, including a random, greedy, uncertainty, diversity, combination of greedy and uncertainty, sampling-based hybrid, and iteration-based hybrid approach. All of these approaches are evaluated and compared using two criteria: (1) the number of identified hits that are selected experiments validated to be responsive, and (2) the performance of the response prediction model trained on the data of selected experiments. The analysis was conducted for 57 drugs and the results show a significant improvement on identifying hits using active learning approaches compared with the random and greedy sampling method. Active learning approaches also show an improvement on response prediction performance for some of the drugs and analysis runs compared with the greedy sampling method.

Toward a universal standard: comparing two methods for standardizing spotted microarray data.

DNA microarray technology has allowed the transcriptome to be studied to a depth that was inconceivable only 10 years ago. Until recently these studies were isolated because, without a universal standard, the results from experiment to experiment and laboratory to laboratory were not directly comparable. For human microarrays, this problem has been addressed by numerous methods, but only two are truly universal. The first method uses genomic DNA as a standard for comparison since it is, by definition, complete and universally available. The second method employs a highly representative total RNA pool such as the one currently available from Stratagene. To determine the advantages and disadvantages of both methods, they were directly compared by hybridization to the University of Texas Southwestern Medical Center's 4000- or 10800-member human cDNA array, using typical microarray techniques. The labeled analytes were 2 microg normal human genomic DNA labeled by nick translation or 20 microg total RNA pool labeled by reverse transcription. The resulting data were then background-subtracted, analyzed, and the number of spots above a background threshold was compared in each sample. Using the McNemar test and a Yate's correction with one degree of freedom, the samples were statistically identical with chi2 = 3.72.

Draft genome sequences of Bordetella holmesii strains from blood (F627) and nasopharynx (H558).

Bordetella holmesii, a human pathogen, can confound the diagnosis of respiratory illness caused by Bordetella pertussis. We present the draft genome sequences of two B. holmesii isolates, one from blood, F627, and one from the nasopharynx, H558. Interestingly, important virulence genes that are present in B. pertussis are not found in B. holmesii.