| Title: | Methods for Analysing Forensic Ancestry Informative Markers |
|---|---|
| Description: | Evaluates likelihood ratio tests for alleged ancestry. Implements the methods of Tvedebrink et al (2018) <doi:10.1016/j.tpb.2017.12.004>. |
| Authors: | Torben Tvedebrink |
| Maintainer: | Torben Tvedebrink <[email protected]> |
| License: | GPL (>= 2) |
| Version: | 0.1.19 |
| Built: | 2024-11-22 05:39:17 UTC |
| Source: | https://github.com/cran/genogeographer |
Shiny application for GenoGeoGrapher
app_genogeo(db_list = NULL, reporting_panel = TRUE)app_genogeo(db_list = NULL, reporting_panel = TRUE)
db_list |
A named list of databases of reference populations.
Each component is expected to be returned from |
reporting_panel |
Logical. Should report generate and download be available after sample analysis. |
Creates the colour scale for the accepted and rejected populations based on z-score and the log likelihood (log P).
bar_colour(df, alpha = 1)bar_colour(df, alpha = 1)
df |
A data.frame with at least three coloums. The first column is the logP, the second logical (z_score accept/reject), the third a unique naming column. |
alpha |
Should the alpha opacity be applied? And what value, 1 = solid, 0 = transparent. |
Plots the estimated profile probabilities in each population. The colour depends on the profiles likelihood and rejection/acceptance (blue/red) based on z-score
error_bar_plot(data)error_bar_plot(data)
data |
The output from the |
A barplot of the log likelihoods for each population with confidence limits
Torben Tvedebrink, [email protected]
df_ <- simulate_pops(pop_n = 20, aims_n = 50) df_db <- pops_to_DB(df_) profile <- random_AIMs_profile(df_db, keep_pop = TRUE) profile$pop[1] # The true population result <- genogeo(profile[,c("locus","x0")], df = df_db) error_bar_plot(result)df_ <- simulate_pops(pop_n = 20, aims_n = 50) df_db <- pops_to_DB(df_) profile <- random_AIMs_profile(df_db, keep_pop = TRUE) profile$pop[1] # The true population result <- genogeo(profile[,c("locus","x0")], df = df_db) error_bar_plot(result)
P-values from Importing Sampling using Exponential tilting
exponent_tilt(x0, x1, n, p_limit = 0.1, B = 500, return_all = FALSE)exponent_tilt(x0, x1, n, p_limit = 0.1, B = 500, return_all = FALSE)
x0 |
Allele count of profile |
x1 |
Population allele count |
n |
Sampled alleles in total in population |
p_limit |
Upper limit to which we use the normal approximation |
B |
An integer specifying the number of importance samples. |
return_all |
Default is FALSE. If TRUE: Returns p-value, standard deviation, and method (and diagnostics). |
The method of importance sampling described in Tvedebrink et al (2018), Section 2.3 is implemented. It relies on exponential tilting of the proposal distribution using in the importance sampling.
If return_all= FALSE the p-value is returned. Otherwise list of elements (see return_all) is returned.
Computes the likelihood ratio test statistics for each population in a database of reference populations.
genogeo(profile, df, CI = 0.95, min_n = 75, grouping = "pop", tilt = FALSE, ...)genogeo(profile, df, CI = 0.95, min_n = 75, grouping = "pop", tilt = FALSE, ...)
profile |
The AIMs profile encoded as returned by the |
df |
The database of reference populations as returned by the |
CI |
The confidence level used to reject or accept the various hypotheses (between 0 and 1). |
min_n |
Minimum number of individuals in each database sample |
grouping |
should |
tilt |
Should exponential titling be used to obtain more accurate $p$-values in the distribution's tail (currently not implemented) |
... |
Further arguments that are passed to other functions |
A tibble containing the $z$-scores, $p$-values etc for each population.
df_ <- simulate_pops(pop_n = 20, aims_n = 50) df_db <- pops_to_DB(df_) profile <- random_AIMs_profile(df_db, keep_pop = TRUE) profile$pop[1] # The true population result <- genogeo(profile[,c("locus","x0")], df = df_db)df_ <- simulate_pops(pop_n = 20, aims_n = 50) df_db <- pops_to_DB(df_) profile <- random_AIMs_profile(df_db, keep_pop = TRUE) profile$pop[1] # The true population result <- genogeo(profile[,c("locus","x0")], df = df_db)
The genogeographer package provides: genogeo()
See ?genogeo
List of markers identified by Kenn Kidd lab.
kidd_locikidd_loci
List of 55 markers
Locus/Marker names
K.K. Kidd et al.Progress toward an efficient panel of SNPs for ancestry inference. Forensic Science International: Genetics 10 (2014) 23–32
For each pair of a specified vector of profiles the likelihood ratios are computed. The list can include all populations in the data or only a subset. We may for inferral purposes restrict to ratios including at least one "accepted" population.
LR_table(result_df, lr_populations = NULL, only_accepted = TRUE, CI = 0.95, digits = NULL, keep_logP = FALSE)LR_table(result_df, lr_populations = NULL, only_accepted = TRUE, CI = 0.95, digits = NULL, keep_logP = FALSE)
result_df |
The output from |
lr_populations |
A vector of population names ( |
only_accepted |
Restrict the ratios to include minimum one accepted population. |
CI |
The level of confidence interval to be computed |
digits |
If rounding of the output should be performed. |
keep_logP |
Logical. Should the logP's be returned in output |
A tibble with numerator and denominator populations with their log10 LR and uncertainty.
Torben Tvedebrink [email protected]
df_ <- simulate_pops(pop_n = 4, aims_n = 50) df_db <- pops_to_DB(df_) profile <- random_AIMs_profile(df_db, keep_pop = TRUE) profile$pop[1] # The true population result <- genogeo(profile[,c("locus","x0")], df = df_db) LR_table(result)df_ <- simulate_pops(pop_n = 4, aims_n = 50) df_db <- pops_to_DB(df_) profile <- random_AIMs_profile(df_db, keep_pop = TRUE) profile$pop[1] # The true population result <- genogeo(profile[,c("locus","x0")], df = df_db) LR_table(result)
List of markers with their main and alternative allele. The markers is the union of Seldin's and Kidd's markers.
main_allelesmain_alleles
List of 164 markers
Locus/Marker names
The main allele (alleles are in lexicographic order)
The other variant
Plots the results from LRT on a map based on lat/lon info in the database.
If no location is found in the data (e.g. using simulte_pops) nothing is plotted.
map_plot(data)map_plot(data)
data |
The output from the |
A map with population z-scores at their geographic origin
Torben Tvedebrink, [email protected]
df_ <- simulate_pops(pop_n = 4, aims_n = 50) df_db <- pops_to_DB(df_) profile <- random_AIMs_profile(df_db, keep_pop = TRUE) profile$pop[1] # The true population result <- genogeo(profile[,c("locus","x0")], df = df_db, min_n = 0) result$lon <- runif(n = 4, min = -125, max = 125) result$lat <- runif(n = 4, min = -50, max = 80) ## Not run: map_plot(result)df_ <- simulate_pops(pop_n = 4, aims_n = 50) df_db <- pops_to_DB(df_) profile <- random_AIMs_profile(df_db, keep_pop = TRUE) profile$pop[1] # The true population result <- genogeo(profile[,c("locus","x0")], df = df_db, min_n = 0) result$lon <- runif(n = 4, min = -125, max = 125) result$lat <- runif(n = 4, min = -50, max = 80) ## Not run: map_plot(result)
Convert the counts from each population over a range of AIMs SNPs q
to observed likelihood ratio test, its mean and variance.
Based on these pre-computed the evaluation of a specific profile is done
using genogeo with the resulting dataframe as df.
pops_to_DB(db, ...)pops_to_DB(db, ...)
db |
A dataframe with columns similar to those of |
... |
Additional arguments passed to |
A tibble with population and locus specific score information
df_ <- simulate_pops(pop_n = 4, aims_n = 50) df_db <- pops_to_DB(df_)df_ <- simulate_pops(pop_n = 4, aims_n = 50) df_db <- pops_to_DB(df_)
Function that compute the genotype probability for each population (rows in df)
profile_AA_x0(AA_profile, df, select = c("locus", "x0"), keep_dropped = FALSE)profile_AA_x0(AA_profile, df, select = c("locus", "x0"), keep_dropped = FALSE)
AA_profile |
A tibble/data.frame with columns 'locus', 'A1' and 'A2' holding the separated version of a genotype, eg. AG -> A1: A, A2: G |
df |
The database with main alleles per locus |
select |
Which columns to return |
keep_dropped |
Logical. Keep the non-matching alleles (compared to 'db') and those with genotype 'NN' |
Compute the z-score (and more) for admixed hypotheses
profile_admixture(x0, df, hyp = NULL, grouping = "meta", return_all = FALSE, calc_logP = TRUE, ...)profile_admixture(x0, df, hyp = NULL, grouping = "meta", return_all = FALSE, calc_logP = TRUE, ...)
x0 |
A data frame/tibble with two columns: 'locus' and 'x0' |
df |
A tibble of reference profiles (as for 'genogeo') |
hyp |
If NULL all levels of 'grouping' is crossed and looped over as pairwise hypotheses. If a single level of 'grouping', this value is crossed with the remaining levels. If vector of two levels this is the only tested hypothesis. |
grouping |
Should the calculations be for meta populations ("meta") or sample populations ("pop")? |
return_all |
Should z-score be returned (FALSE) or all locus results (TRUE)? |
calc_logP |
Should log P(Geno|Hyp) be calculated (TRUE) or not (FALSE)? |
... |
additional arguments passed on to other functions |
A tibble of z-scores, or a list of pairwise results if 'return_all = TRUE'
Use the information from pops_to_DB to simulate a profile from a random or given population.
The sampling is done with respect to the null hypothesis, such that the total count is adjusted accordingly.
For further details see Tvedebrink et al (2018), Section 3.1 (Simulations).
random_AIMs_profile(df, grouping = "pop", population = NULL, n = FALSE, keep_pop = FALSE)random_AIMs_profile(df, grouping = "pop", population = NULL, n = FALSE, keep_pop = FALSE)
df |
Database of reference profiles as returned by |
grouping |
Simualte from |
population |
The population to sample from. If NULL chosen at random. |
n |
Use numbers of samples as weights to choose the population randomly |
keep_pop |
Keep information on population |
Torben Tvedebrink [email protected]
List of markers identified by Seldin lab.
seldin_lociseldin_loci
List of 122 markers
Locus/Marker names
Kosoy et al. Ancestry Informative Marker Sets for Determining Continental Origin and Admixture Proportions in Common Populations in America. HUMAN MUTATION, Vol. 30, No. 1, 69–78, 2009.
Simulate random populations
simulate_pops(pop_n = 100, pop_names = NULL, pop_totals = NULL, aims_n = 50, aims_names = NULL)simulate_pops(pop_n = 100, pop_names = NULL, pop_totals = NULL, aims_n = 50, aims_names = NULL)
pop_n |
Number of populations to simulate |
pop_names |
Their names. If NULL: The names are "pop_001" through "pop_pop_n" |
pop_totals |
How many observations/sampled individuals per population. If one number this is used as parameter in a Poisson distribution |
aims_n |
Number of AIMs |
aims_names |
Their names. If NULL: The names are "rs_001" through "rs_aims_n" |
Torben Tvedebrink [email protected]