deletions | additions       diff --git a/A_bootstrap_sub_sampling_method__.tex b/A_bootstrap_sub_sampling_method__.tex  index b4c7cec..a88e1d4 100644  --- a/A_bootstrap_sub_sampling_method__.tex  +++ b/A_bootstrap_sub_sampling_method__.tex 
...
A bootstrap sub-sampling method was used to reduce the initial panel  of 500k SNPs down to a set that consistently produced associations on  all bootstrap sub-samples. Sub-samples of the Type 1 Diabetes (T1D)  cases and National Blood Service (NBS) controls were used to estimate  the in-group variance of association statistics throughout the genome.  Markers that were informative and had low variance were selected as  candidate markers for a minimal informative set of 45 markers.  The final refinement step tested sets of SNPs in combination, rather  than single SNPs alone, in the hope that those sets would be able to  capture a wider range of genetic variation than any single marker (or  combination of data for single markers alone) could provide. Once a  suitable SNP set was found, that set was tested in the validation  group to confirm the utility of the set for distinguishing T1D cases  from NBS controls.  \subsection{Genotyping and Filtering of Individuals and SNPs}  \label{sec:meth-summ-genotyping-filtering} 
...
scores ($\chi^2 > 100$) were found on chromosomes 3, 6, 12, 16, and  22, but the scores were only consistently high for a region of about  10Mb in the middle of the short arm of chromosome 6 (30-40Mb from the  $5^\prime$ end of the forward strand).