Microsatellites or short tandem repeats are polymorphic loci present in the human genome. They consist of repeating units of 1-6 base pairs. The repeat numbers of these microsatellites are analyzed at four loci, TPOX, PK211, D13, and CSF. The short tandem repeats can be considered to be alleles and can be passed down to next generation. When mosquitoes feed from humans, they also play the role of sampling these alleles. The blood meal of mosquitoes was analyzed and the microsatellites at four loci were extracted from white blood cell DNA, amplified by PCR, and eventually detected by electrophoresis. The number of different types of alleles in the blood meal can be used to set a lower bound on the numbers of blood meals. For instance, if there are three types of alleles at TPOX in the same blood meal, it means that the mosquito must have bitten at least two individuals since each feeding can contribute at least two alleles. The multiple blood meal probability is important information. Because two feedings are required to transmit the parasite from one host to another combined with mosquito life span it could tell us how fast the disease is delivered by mosquitoes. If all mosquitoes only feed once or multiple bites always feed from a single, then mosquitoes could not transmit disease from one host to another.
Data Source All microsatellites data from 2005-2007 were recruited from
Douglas Norris' laboratory (Johns Hopkins University, School of Public Health, Department of Molecular Microbiology Immunology)
Lower bound method Assume there is a human population with only five types of the genotypes AC, AB, BC, AA, BB, and CC at locus TPOX. There are also some mosquitoes captured and their blood meals are analyzed. In the first blood meal, only A and B alleles are detected. There are two deductions for this phenomenon. First, this mosquito feeds from a single with AB once or more than once. Second, this mosquito feed from several individual with the identical genotype AB. However, one could not be sure how many individuals’ blood meal are involved. To be scientifically conservative, one can only say, “this mosquito at least feeds from a single ” or “the lower bound on this blood meal number is one” In the second case, the blood meal is detected to have A, B, and C. It is known that this mosquito at least fed from two individuals with the possible genotype combinations, AB and AC, BC and AC, or BC and AB. And one can say, “the lower bound on this blood meal number is two”, but it is always possible to have more individuals involved. In the experimental data, microsatellite alleles of the four loci are measured and the maximum allele number is taken to rule out the underestimation of lower bound on blood meal number. Therefore, the relationship between the maximum allele number and the lower bound on blood meal number was formulated and this method is named “lower bound method”. The following shows how lower bound method works.
Lower bound of blood meal
number = round (maximum allele number detected/2)
the Lower bound on blood meal number
Maximum allele number detected in Locus
1
1
1
2
2
3
2
4
3
5
3
6
4
7
4
8
5
9
5
10
Resulting blood meal probabilities and theoretical blood meal probabilities
Theoretical blood meal probabilities were used to simulate how frequently the mosquito feed from humans. Resulting blood meal probability is the normalized form of the lower bound on blood meal number. For instance, if 20 mosquitoes’ lower bounds on the blood-meal number are estimated to be two and 80 mosquitoes’ lower bounds on blood-meal number are estimated to be one, it is known that the resulting multiple blood-meal probability is calculated to be 20%.
Basic simulation flow
With allele frequencies provided from the laboratory 2005-2007 data, under Hardy Weinberg equilibrium assumption, next generation humans with the certain population size was first created. This human group was further fed to mosquito population with theoretical blood meal probabilities to generate blood meal data set. With PCR detection probabilities obtained from the same laboratory data, PCR simulation assigned bad PCR reaction product to each sample to generate the imperfect blood meal data (the detail will be discussed in the section” Detection Probabilities"). Those mosquito individuals with bad PCR reaction product were finally eliminated. The mosquito samples left must have four different loci (TPOX, PK211, D13, and CSF) and each locus has at least one type of microsatellite allele. Then the lower bound on blood meal number will be calculated by lower bound method. The lower bound on blood meal number is further normalized to be resulting blood meal probability.