Research

Sam's primary interest is engineering software solutions to biological problems.

Sam Danziger�s research focused on improving the accuracy and effectiveness of computational machine learning systems using computer models and in silico predictions applied to in vivo and in vitro experimentation for biomedical projects. Specifically, he developed novel computational active learning methods in conjunction with homology modeling and feature extraction techniques for the p53 cancer rescue mutant problem. p53 is a prominent protein in the human cancer prevention pathway, and is encoded by the TP53 gene. Certain p53 cancer mutants regularly occur in human tumors and are inactive in biological assays. When some p53 cancer mutants are given additional second site rescue mutations, the p53 is reactivated. It is theorized that understanding the p53 rescue mechanism can aid in the design of small molecule drugs that mimic the rescue effect. Sam used his initial in silico predictions to find p53 rescue mutants that are active in vivo and in vitro and to predict the behavior of those mutants with up to 77% accuracy. His subsequent research developed novel active learning strategies for most rapidly and efficiently building an accurate classifier using the fewest number of expensive data points. His most recent research developed a novel active learning strategy specialized to find positive examples quickly. The p53 prediction algorithms were extended to detect cancer rescue hot-spot domains with proficiency similar to that of an expert biologist. These algorithms found p53 cancer rescue mutants in silico 33% faster than regular active learning, and rescued for the first time the previously unrescuable cancer mutant P152L.

One year of Sam�s research time was spent in Dr. Brachmann�s lab expressing mutant p53 in yeast, during which time he also developed software to aid in the design of mutant oligonucleotides and primers. Sam�s computer models were also used by Dr. Ruslan Aphasizhev to design mutants of the RNA editing proteins terminal uridylyltransferases TUT4 and RET2.

Click here for the UCI p53 project page

Papers

• Bernabeu M., Danziger S.A., Avrila M., Vazb M., Babar P., Braziera A.J., Herricks T.E., Maki J.N., Pereira L., Mascarenhas A., Gomes E., Chery L., Aitchison J.D., Rathod P.K., and Smith J.D. (2016) Severe adult malaria is associated with specific PfEMP1 adhesion types and high parasite biomass, Under Embargo, TBD, (link)


• Wang Z., Danziger S.A., Heavner B.D., Ma S., Smith J.J., Li S., Herricks T., Simeonidis E., Baliga N.S., Aitchison J.D., and Price N.D. (2016) Combining inferred regulatory and reconstructed metabolic networks enhances phenotype prediction in yeast, Under Revision, TBD, (link)


• Danziger S.A., Miller L.R., Signh K, Peskind E.R., Li G., Lipshutz R., Aitchison J.D., and Smith J.J. (2016) An indicator cell assay for blood-based diagnostics, Under Review, TBD, (link)


• Jung S, Danziger S.A., Panchaud A., von Haller P., Aitchison J.D., and Goodlett D.R. (2015) Systematic Analysis of Yeast Proteome Reveals Peptide Detectability Factors for Mass Spectrometry, Journal of Proteomics & Bioinformatics, 8(10), 231-239 (link)


• Danziger, S.A., Reiss, D.J., Ratushny, A.V., Smith, J.J., Plaisier, C.L., Aitchison, J.D., and Baliga, N.S. (2015) Bicluster Sampled Coherence Metric (BSCM) provides an accurate environmental context for phenotype predictions, BMC Systems Biology, 9, S1 (link)


• Finney O.C.*, Danziger S.A.*, Molina D.M., Vignali M., Takagi A., Ji Mi., Stanisic D.I., Siba P.M., Liang X., Aitchison J.D., Mueller I., Gardner M.J., and Wang R. (2014) Predicting anti-disease immunity using proteome arrays and sera from children naturally exposed to malaria, Molecular & Cellular Proteomics, 13(10), 2646-2660 (link)


• Danziger, S.A., Ratushny, A.V., Smith, J.J., Saleem, R.A., Wan, Y., Arens, C.E., Armstrong, A.M., Sitko, K., Chen,W.-M., Chiang, J.-H., Reiss, D.J., Baliga, N.S. and Aitchison, J.D. (2014) Molecular Mechanisms of System Responses to Novel Stimuli are Predictable from Public Data, Nucleic Acids Research, 42(3), 1442-1460 (link)


• Chen,W.-M.*, Danziger,S.A*., Chiang,J.-H. and Aitchison,J.D. (2013) PhosphoChain: a novel algorithm to predict kinase and phosphatase networks from high-throughput expression data, Bioinformatics , 29(19), 2435-2444 (link)


• Cooney,L.A., Gupta,M., Thomas,S., Mikolajczak,S., Choi,K.Y., Gibson,C., Jang,I.K., Danziger,S.A., Aitchison,J., Gardner,M.J., et al. (2013) Short-Lived Effector CD8 T Cells Induced by Genetically Attenuated Malaria Parasite Vaccination Express CD11c, Infection and Immunity, 81(11), 4171-4181 (link)


• Wan,Y., Zuo,X., Zhuo,Y., Zhu,M., Danziger,S.A. and Zhou,Z. (2013) The functional role of SUMO E3 ligase Mms21p in the maintenance of subtelomeric silencing in budding yeast, Biochemical and Biophysical Research Communications, 438(4), 746-752 (link)


• Baronio, R., Danziger, S.A., Hall, L.V., Salmon, K., Hatfield, G.W., Lathrop, R.H., and Kaiser, P. (2010) All-codon scanning identifies p53 cancer rescue mutations, Nucleic Acids Research, 38(20), 7079-7088 (link)


• Danziger, S.A., Baronio, R., Ho, L., Hall, L., Salmon, K., Hatfield, G.W., Kaiser, P., and Lathrop, R.H. (2009) Predicting Positive p53 Cancer Rescue Regions Using Most Informative Positive (MIP) Active Learning, PLOS Computational Biology, 5(9), e1000498 (link)


• Danziger, S.A., Zeng, J., Wang, Y., Brachmann, R.K. and Lathrop, R.H. (2007) Choosing where to look next in a mutation sequence space: Active Learning of informative p53 cancer rescue mutants, Bioinformatics, 23(13), 104-114. (link, prerelease)


• Danziger, S.A., Swamidass, S.J., Zeng, J., Dearth, L.R., Lu, Q., Chen, J.H., Cheng, J., Hoang, V.P., Saigo, H., Luo, R., Baldi, P., Brachmann, R.K. and Lathrop, R.H. (2006) Functional census of mutation sequence spaces: the example of p53 cancer rescue mutants, IEEE/ACM transactions on computational biology and bioinformatics / IEEE, ACM, 3, 114-125. (link, code and data)


• Danziger, S. A. and Anderson, P.G. (2002) Next Generation Optical Character Recognition Using the Polynomial Method Rochester Institute of Technology. (Originally written as a disclosure of patentable invention. Released in 2003) (pdf)


• Danziger, S.A., Salehi-Amiri S.F., Baronio, R., Hall, L., Hatfield, G.W., Kaiser, P., and Lathrop, R.H. (2009) Better Active Learning by Choosing the Most Informative Positive (MIP) Examples, Under Revision

Wetlab Protocols

While Sam background is Electrical Engineering and Computer Science, he recently had the opportunity to work in Dr. Brachmann's Molecular Biology lab. To ease the transition, he assembled the following:

• How to make mutants: A detailed step by step guide of the lab protocols necessary to make p53 mutants and express them in yeast (in progress). (PDF)

Posters and Presentations (A Sampling)

• Roberta Baronio, Linda V. Hall, Lydia Ho, Kirsty A. Salmon, G. Wesley Hatfield, Peter Kaiser and Richard H. Lathrop (2009) Machine Learning discovers p53 cancer rescue regions. Center for Machine Learning & Intelligent System - Seminar 2008-2009 (PPT).


• Danziger, S.A. Zeng, J., Brachmann, R.K. and Lathrop, R.H. (2007) Choosing Where to Look Next in a Mutation Sequence Space: Active Learning of Informative p53 Cancer Rescue Mutants. ISMB/ECCB 2007 (PPT).


• Danziger, S.A. (2007) Helping to Cure Cancer: Computer Science and Biology. Ask-A-Scientist Night: Hillview Elementary (PPT).


• Danziger, S.A., Zeng, J., Brachmann, R.K. and Lathrop, R.H. (2006) In Silico Protein Behavior: Predicting the Activity of p53 Tumor Suppressor Protein Mutants Using Features Derived From Homology Modeling. 9th Annual UCI Cancer Center Retreat (PPT).


• Danziger, S.A, Brachmann, R.K. and Lathrop, R.H. (2005) Predicting Mutant Protein Function Using Computer Models. Samueli Scholars Luncheon (TIF).


• Danziger, S.A, Brachmann, R.K. and Lathrop, R.H. (2005) Predicting Mutant Protein Function Using Computer Models. 2005 National Library of Medicine Informatics Training Programs Conference (TIF).