hapsburg.PackagesSupport.fit_ne

Class the fits Ne with a MLE framework. Standard error either from bootstrap over individuals, or from likelihood curve (kind of assumping ROH are independent, which is not fully correct) @Author: Harald Ringbauer, 2020

Classes

MLE_ROH_Ne

Class for MLE estimation of block sharing

Functions

`ind_roh_from_pdf`([ne, bin_range, nbins, output])	Create ROH for single individual. Everything measured in Morgan.
`inds_roh_from_pdf`([n_ind, ne, bin_range, nbins, ...])	Create ROH for single individual. Everything measured in Morgan.
`load_roh_vec`([iids, base_path, suffix])	Load and return ROH length vector
`get_default_res`()	REturn default line of results dataframe.

Module Contents

class hapsburg.PackagesSupport.fit_ne.MLE_ROH_Ne(endog=[], exog=[], start_params=[1000], min_len=4, max_len=20.0, error_model=False, output=False, chr_lgts=[], **kwds)

Bases: statsmodels.base.model.GenericLikelihoodModel

Class for MLE estimation of block sharing between populations with modeled error Bins into length blocks and models block-sharing between populations as indep. Poisson Operates in cM unit space (not Morgan!)

bin_width = 0.1

mid_bins = []

chr_lgts = []

fp_rate = []

theoretical_shr = []

trans_mat

full_shr_pr = []

density_fun = 0

start_params = []

error_model = True

estimates = []

summary = 0

output = True

initialize_ll_model(**kwds): Function to initialize LL model

set_params(recalculate_bins=False, **kwargs): Set the Parameters. Takes keyworded arguments

fit(start_params=None, maxiter=10000, maxfun=5000, **kwds): Fit Parameters via fit method of statsmodels

fit_ll_profile(ns=[], level=1.92, num=1000): Fit Parameters via likelihood profile. Only works for 1D search. Return ML, lower, upper CI values ns: Array of 2Ne values to test. level: loglikelihood difference to use for CI values.

loglikeobs(params): Return vector of log likelihoods for every observation. (here pairs of pops)

ll_individual(roh, params): Log likelihood function for every raw of data (sharing between countries). Return log likelihood.

create_bins(): Creates the bins according to parameters in Class. Also set the false positive rate

calculate_thr_shr(params): Calculates the expected Bessel-Decay per bin

calculate_trans_mat(): Calculate the transition matrix from true estimated to observed values for block sharing.

calculate_full_bin_prob(): Calculate the full probablities per bin

get_bl_shr_interval(interval, r, params=[]): Return the estimated block-sharing under the model in interval given distance r. For this use all the bins intersecting the interval and average. Assumes r is array and return array

print_block_nr(): Calculate and print the Nr of analyzed Blocks of right Length.

ci_li(ns=[], level=1.92, update=True): Get confidence Intervall based on likelihood ration test. Return lower, upper Ne supported by ns: N values to test. level: loglikelihood difference to use. update: Whether to update Confidence Intervals in the fit

block_shr_density(l, params): Calculate and print the Nr of analyzed Blocks of right Length.

get_summ_as_df(summary): Transforms statsmodels summary to pandas dataframe. If no summary given use the last saved one Return that dataframe

hapsburg.PackagesSupport.fit_ne.ind_roh_from_pdf(ne=500, bin_range=[0.04, 0.5], nbins=100000, output=False): Create ROH for single individual. Everything measured in Morgan. Return array of ROH lengths, in Morgan

hapsburg.PackagesSupport.fit_ne.inds_roh_from_pdf(n_ind=5, ne=500, bin_range=[0.04, 0.5], nbins=100000, output=False, cm=True): Create ROH for single individual. Everything measured in Morgan. Return array of ROH lengths, in Morgan. cm: If true, output in centimorgan

hapsburg.PackagesSupport.fit_ne.load_roh_vec(iids=[], base_path='./output/roh/', suffix='_roh_full.csv'): Load and return ROH length vector

hapsburg.PackagesSupport.fit_ne.get_default_res(): REturn default line of results dataframe. Used for unfittable scenarios