## Construction of consensus sequences for natural populations
To construct population specific consensus sequences, we used ANGSD ([Korneliussen et al. 2014](http://www.biomedcentral.com/1471-2105/15/356/abstract)) and the 'doFasta 2' option.
To construct population specific consensus sequences, we used 'ANGSD' ([Korneliussen et al. 2014](http://www.biomedcentral.com/1471-2105/15/356/abstract)) and the 'doFasta 2' option.
##### CONSENSUS FASTA
In addition to the 'doFasta 2' optin, which extracts the most common nucleotide, other filters like 'minQ' and coverage filters like 'setMinDepthInd' and 'setMaxDepthInd' were applied.
CONSENSUS FASTA example for the population _Mus musculus musculus AFG_:
CONSENSUS FASTA example for the population _Mus musculus musculus AFG_
```
#example for populaton Mmm_AFG for chr1
...
...
@@ -101,13 +101,13 @@ All CONSENSUS FASTA files can be obtained from:
## Construction of individual sequences using IUPAC code
To construct individual sequences, we used ANGSD and the 'doFasta 4' option. For this option, after all filters have been applied, all multi-allelic sites will be encoded as IUPAC code.
To construct individual sequences, we used 'ANGSD' and the 'doFasta 4' option. For this option, after all filters have been applied, all multi-allelic sites will be encoded as IUPAC code.
##### INDIVIDUAL FASTA
In addition to the 'doFasta 4' optin, which extracts IUPAC code, other filters like 'minQ' and coverage filters like 'setMinDepthInd' and 'setMaxDepthInd' were applied.
CONSENSUS FASTA example for the population _Mus musculus musculus AFG_:
CONSENSUS FASTA example for the population _Mus musculus musculus AFG_
```
#example for one Mus musculus musculus individual - AFG1 - 396 for chr1
where $d_{XO_{w_{i}}} and $d_{XZ_{w_{i}}} are defined as the average Kimura's 2-parameter sequence distance ([Kimura 1980](https://www.ncbi.nlm.nih.gov/pubmed/7463489)) between the corresponding two populations calculated with the function 'dist.dna' of the of the R package 'ape' ([Paradis et al. 2004](https://academic.oup.com/bioinformatics/article/20/2/289/204981/APE-Analyses-of-Phylogenetics-and-Evolution-in-R)) using the model 'K80'. Prior the calculation of dK80 all sites with missing data within the specified window ($w_{i}$) and the specified populations were removed across the whole quartet with the 'Biostrings' R package ([Pages et al. 2009](https://bioconductor.org/packages/release/bioc/html/Biostrings.html)).
example for chromosome 1 for the dK80 calculation for the quartet [X]: _Mus musculus domesticus FRA; [Y]: _Mus musculus domesticus GER; [Z]: _Mus musculus domesticus IRA; [O]: _Mus musculus musculus AFG_:
example for chromosome 1 for the dK80 calculation for the quartet [X]: _Mus musculus domesticus FRA; [Y]: _Mus musculus domesticus GER; [Z]: _Mus musculus domesticus IRA; [O]: _Mus musculus musculus AFG_
```
#example for the quartet [X]: FRA; [Y]: GER; [Z]: IRA; [O]: AFG
Phylogenetic trees per 25kbp windows were calculated with the 'bionjs' function of the R package 'ape' ([Paradis et al. 2004](https://academic.oup.com/bioinformatics/article/20/2/289/204981/APE-Analyses-of-Phylogenetics-and-Evolution-in-R)) and used as input files for 'twisst' ([Martin and Belleghem 2017](http://www.genetics.org/content/early/2017/03/21/genetics.116.194720)). The 'bionjs' function can reconstruct a phylogenetic tree from a distance matrix with possibly missing values. We developed an R package (https://github.com/kullrich/distIUPAC) to calculate pairwise 'literal distances' using IUPAC encoded nucleotide sequences. Briefly, only bi-allelic nucleotide states are considered for the pairwise distance calculation and all tri-nucleotide states, gaps or missing data are discarded. The distance score matrix applied is based on ([Chang et al. 2017](https://www.ncbi.nlm.nih.gov/pubmed/28819774)), whereby two homologous SNPs yield 0 if they were identical or identically heterozygous, 1 if different, and 0.5 if one of them was heterozygous. Windows with complete deletion sites (missing in all individuals) higher than 50\% were removed. The resulting pairwise distance matrix per window were used to calculate phylogenetic trees which were further used as input trees for 'twisst'.
example for chromosome 3 for the tree calculations
```
#example for chromosome 3
#change the bottom part of the script 'get_bionj_trees_for_twisst.r' for each chromosome
FASTA sequence construction and dK80 calculation was performed as described above.
## Mappability mm10 genome
To construct a mappability track we used the software GEM ([Derrien et al. 2012](https://www.ncbi.nlm.nih.gov/pubmed/22276185)) with follofing options on the mm10 genome.
K-MER LENGTH: 100
APPROXIMATION THRESHOLD: 7
MAX MISMATCHES: 4
MAX ERRORS: 4
MAX BIG INDEL LENGTH: 15
MIN MATCHED BASES: 80
STRATA AFTER BEST: 1
The bigWig track file (http://wwwuser.gwdg.de/~evolbio/evolgen/wildmouse/introgression/mm10.gem.mappability.25kbp.sorted.bed.bw) represents the mean values for 25kbp windows.
