BTRY 6940 Graduate Special Topics in Biometry and Statistics

Instructor: Jim Booth
Time: 2:15 PM on Fridays
Place: 1181 Comstock Hall
1 credit

The goal of this class is to review the current state of the literature on statistical and computational methods for network and pathway discovery in biological systems.

The class will meet one hour per week. Students signed up for credit will be expected to do a presentation (say 30 minutes) on a relevant paper or set of related papers.

Ordered List by Date: Presenter :: Topic ::: Background Readings

1. September 12 2014 : Kelson Zawack :: The graphical lasso and the nonparanormal skeptic model with applications to some mass spectrometry data (PDF slides 2068KB) ::: Methodology papers for graphical lasso and nonparanormal skeptic.
2. September 19 2014 : Irina Gaynanova :: Introduction to Lasso and Graphical Lasso (PDF slides 244KB) ::: See lasso for a simple explanation and the original paper.
3. September 26 2014 : David Sinclair :: Introduction to Networks in Brain Imaging (PDF slides 508KB) ::: See Network diffusion accurately models the relationship between structural and functional brain connectivity networks and Network modelling methods for FMRI.
4. October 03 2014 : James Booth :: Introduction to Factor Analysis and eQTL (PDF slides 1768KB) ::: See eQTL/factor analysis paper by Gao et al..
5. October 08 2014 : Adam Rothman :: Properties of optimizations used in penalized Gaussian likelihood inverse covariance matrix estimation (PDF slides 432KB).
6. October 10 2014 : Christian Mueller (Center for Genomics and Systems Biology, NYU) :: Inference of microbial ecological interaction networks with SParse InversE Covariance selection for Ecologial ASsociation Inference [SPIEC-EASI] (PDF slides 19MB) ::: See http://arxiv.org/pdf/1408.4158v2.pdf.
7. October 17 2014 : Daqian Sun :: Large Metabolic Networks (PDF slides 296KB) ::: See Metabolic Network paper by Barabasi Lab and the Chapter on Preferential Attachment in http://www.win.tue.nl/~rhofstad/NotesRGCN.pdf.
8. October 31 2014 : Yrjo Tapio Grohn (Cornell College of Veterinary Medicine) :: Progression to multi-scale models and the application to food system intevention strategies (in press in Preventitive Veterinary Medicine)
9. November 5 2014 : Rachael Hageman Blair (University at Buffalo) :: Belief Propagation in Genotype-Phenotype Networks (PDF slides 4.9MB) : See Graduate Lectures in Graphical Models and Causal Inference at Steffen L. Lauritzen's page http://www.stats.ox.ac.uk/~steffen/teaching/index.htm and his Wald lectures (2012 Istanbul) for theoretical background on Rachel's talk.
10. November 14 2014 : Raazesh Sainudiin (University of Canterbury, Christchurch, New Zealand; on Sabbatical at Dept.~of Mathematics, Cornell) :: Some Experiments in Phylodynamic Epidemiology (PDF scan of white-board lecture notes 3.0MB) ::: See Phylogenetic tree shapes resolve disease transmission patterns, Caroline Colijn and Jennifer Gardy, Evolution Medicine and Public Health Advances, OUP, June 9, 2014 for applied statistical/public-heath motivation of the lecture-discussion and see Unifying the Epidemiological and Evolutionary Dynamics of Pathogens for one of the "founding paper(s)". For a current-affairs perspective on how the host-contact network structure is based on family bonds derived from pedigrees see For a Liberian Family, Ebola Turns Loving Care Into Deadly Risk, By Norimitsu Onishinov on November 13, 2014, NY Times. For a recent work on notions of ancestries in a population's pedigree within which cytoplasmic parasites such as Wolbachia are transmitted see Ancestries of a Recombining Diploid Population, Raazesh Sainudiin, Bhalchandra Thatte, and Amandine Véber, UCDMS Research Report 2014/3, 42 pages, 2014.

Some Stats Seminars of Possible Interest

• November 12 2014 : R. Dennis Cook (University of Minesota) :: Envelopes; Methods for Efficient Estimation in Multivariate Statistics (PDF slides 5.2MB). Further reading on their current initialization algorithm to search over "optimally enveloping Grassmanian" is described in section 4 of http://arxiv.org/abs/1403.4138 and its output is root-n consistent, and it can also be used as a stand-alone method.

Background reading on Gaussian Graphical Models (from Lingzhou Xue via Kelson Zawack)

• covariacne selection: Dempster, '72;
• testing approach: Drton and Perlman, '04, '07, '08;
• neighborhood approach: Meinshausen and Buhlmann, '06; Yuan, '10;
• GLasso and its variant: Yuan and Lin, '07; Friedman et al, '08; Banerjee et al, '08; Rothman et al, '08; Ravikumar et al, '08; Lam and Fan, '09; Peng et al, '09; Scheinberg et al, '10; Witten et al, '11;
• constrained minimization: Cai et al, '11;
• Gaussian copula graphical model: Xue and Zou, '12; Liu et al., '12;
• latent variable Gaussian graphical model: Chandrasekaran et al, '12; Ma, Xue and Zou, '13;

Unordered List of Potential relevant Topics

We will decide at the first meeting about other presenters and topics. Here are several papers that might be interesting to examine in detail. If you have suggestions for other topics send a link with a brief description in plan ASCII text to my email address.

• Network-based stratification of cancer patients:
  Chuang, HY, Lee, E, Liu, YT, Lee, D, and Ideker, T. Network-based classification of breast cancer metastasis. Mol Syst Biol. 3:140 (2007)
  Dynamic modularity in protein interaction networks predicts breast cancer outcome. Ian W Taylor, Rune Linding, David Warde-Farley, Yongmei Liu,Catia Pesquita, Daniel Faria, Shelley Bull, Tony Pawson, Quaid Morris, Jeffrey L Wrana. Nature Biotechnology 03/2009; 27(2):199-204.
  Hofree M, Shen JP, Carter H, Gross A, Ideker T. Network-based stratification of tumor mutations. Nat Methods 10(11):1108-15 (2013)
• Network modelling methods for FMRI
• Network Genome-wide Association Studies (GWAS):
  Network-based analysis of genome wide association data provides novel candidate genes for lipid and lipoprotein traits. Sharma A1, Gulbahce N, Pevzner SJ, Menche J, Ladenvall C, Folkersen L, Eriksson P, Orho-Melander M, Barabási AL. Mol Cell Proteomics. 2013 Nov;12(11):3398-408.
• Graphical model GWAS:
  Bayesian Graphical Models for Genomewide Association Studies, Claudio J. Verzilli, Nigel Stallard, and John C. Whittaker, Am J Hum Genet. Jul 2006; 79(1): 100–112.
  A novel bayesian graphical model for genome-wide multi-SNP association mapping. Yu Zhang. Genetic Epidemiology Volume 36, Issue 1, pages 36–47, January 2012
• Network decomposition:
  Network link prediction by global silencing of indirect correlations, Baruch Barzela and Albert-László Barabási, Nature Biotechnology 31, 720–725 (2013)
  Network deconvolution as a general method to distinguish direct dependencies in networks. Soheil Feizi, Daniel Marbach, Muriel Médard and Manolis Kellis, Nature Biotechnology 31, 726–733 (2013)
• Lange et al (2014), Annu. Rev. Stat. Appl. http://www.annualreviews.org/doi/pdf/10.1146/annurev-statistics-022513-115638
• Gao et al (2014), Bioinformatics http://bioinformatics.oxfordjournals.org/content/30/3/369.full.pdf+html?sid=61d57e03-b524-4aeb-b6b7-6243d2e63c2c
• Battle et al (2014), Genome Research http://genome.cshlp.org/content/24/1/14.full.pdf+html
• http://www.biostat.washington.edu/~dwitten/Papers/pmd.pdf

Coordinated by Raazesh Sainudiin
with partial support from:
2014 Sabbatical grant from College of Engineering, University of Canterbury, Christchurch, New Zealand and
a Visiting Scholarship to Department of Mathematics, Cornell University, Ithaca, New York