New models and methods for interpreting molecular epidemiological data

A major focus of our group is understanding the transmission of diseases such as tuberculosis by extracting information from molecular data.

The infinite alleles model and Ewens’s sampling theory can be used to understand the distribution of sizes of genotype clusters in bacterial isolates [1]. See Figure 1.

We have estimated the reproductive value of a pathogen by using a stochastic model of transmission and mutation with approximate Bayesian computation [2].

We are developing new computational methods to estimate parameters [3] and applying these methods to study the transmission of drug resistance in tuberculosis.

We are formulating models of marker evolution based on typing technologies such as spoligotyping. These models can be used to develop methods to visualise the diversity of strains such as in samples of tuberculosis isolates. See Figure 2.

Information about the extent to which each genotype has mutated allows inference about the relative population growth rates among different strains [4].

We are building a web-accessible computational toolkit and database for analysing tuberculosis genotypes based on the spoligotyping technology.

Figure 1. Left: estimates of diversity parameter θ from 128 tuberculosis data sets, 75 data sets genotyped by IS6110 (diamonds), 53 data sets by spoligotyping ("+"). Upper and lower dashed lines represent first and third quartiles respectively. The IS6110 patterns mutate faster than spoligotypes. Right: relative mutation rates of markers given by the ratio of estimated θ values for each marker.

Evolutionary genomics and models of host pathogen systems

The genome of Shigella flexneri has an unusually large number of insertion sequences, which are likely to be involved in the adaptive evolution of the bacterium within the human host [5].

We are investigating the influence of hotspots, short nucleotide motifs targeted by somatic hypermutations, on antibody sequence evolution and the ultimate effect of these on the immune response.

Models of protein degradation by proteasomes predict the distribution of peptides recognised by the immune system. These models contribute to an understanding of the interaction between pathogen and MHC variation [6].

Mathematical modelling of within-cell dynamics of positive single-stranded RNA viruses such as the SARS virus and Hepatitis C Virus [7].

We are interested in the interaction between pathogens and behavioural traits of hosts that influence the outcome of the infection.

Figure 2. Left: Sample of spoligotyped tuberculosis isolates from Cuba. Spoligotype patterns evolve through the deletion of spacer sequences that cannot be recovered. Right: A spoligoforest shows the relationships among patterns. Ambiguities of ancestry (shown with red edges) were resolved using a mutation model. For instance, spoligotype 43 is inferred to have descended from spoligotype 14 among 18 possible ancestors. [Data from Diaz, et al. Int J Tuberc Lung Dis. 1998 Sep;2(9):743-50]

Recent collaborators

SA Sisson (University of New South Wales)
Y Fan (University of New South Wales)
R Lan (University of New South Wales)
AR Francis (University of Western Sydney)

Selected publications

[1] Luciani F, Francis AR, Tanaka MM, Interpreting genotype cluster sizes of Mycobacterium tuberculosis isolates typed with IS6110 and spoligotyping. Emerging Infectious Diseases, Submitted.

[2] Tanaka, MM, Francis AR, Luciani F, Sisson SA, Using approximate Bayesian computation to estimate tuberculosis transmission parameters from genotype data. Genetics. 2006 Jul; 173(3) : 1511-20.

[3] Sisson SA, Fan Y, Tanaka MM, Sequential Monte Carlo without likelihoods. Proc Natl Acad Sci U S A. 2007 Feb 6;104(6):1760-5.

[4] Tanaka MM, Francis AR, Detecting emerging strains of tuberculosis by using spoligotypes. Proc Natl Acad Sci U S A. 2006 Oct 10;103(41):15266-71.

[5] Zaghloul L, Tang C, Chin HY, Bek EJ, Lan R, Tanaka MM, The distribution of insertion sequences in the genome of Shigella flexneri strain 2457T. PloS One, Submitted.

[6] Luciani F, Nielsen M, Kesmir C, Di Napoli M, Theoretical models for proteasome function: predictive methods to understand ubiquitin protein system in neurodegenerative diseases, in The UPS in the nervous system: from physiology to pathology, ed. M Di Napoli and C Wojcik, pp. 1-36, Nova Science Publishers, Inc: 2006.

[7] Regoes RR, Crotty S, Antia R, Tanaka MM, Optimal replication of poliovirus within cells. Am Nat. 2005 Mar;165(3):364-73.