ska distance - simonrharris/SKA GitHub Wiki
The distance subcommand allows calculation of pairwise distances between and single-linkage clustering of samples in split kmer files based on user-defined SNP and identity cutoffs.
The clustering method employed is very simple. Samples are clustered if they meet the requirements of both of the following
- If the number of SNPs between them is less than the SNP cutoff set with the -s option [Default = 20] and
- They meet the identity cutoff set with the -i option [Default = 0.9]. I.e. they share at least this proportion of the total number split kmers in the file with fewer kmers.
| Column | Description |
|---|---|
| Sample 1 | The name of the first sample being compared |
| Sample 2 | The name of the first sample being compared |
| Matches | Number of split kmers found in both samples where the middle base is an A, C, G or T and matches between samples |
| Mismatches | Number of split kmers found in only one of the samples |
| Jaccard Index | Ratio of split kmers found in both samples to the total found in the two samples: matches/(matches+mismatches) |
| Mash-like distance | A distance based on the Mash distance calculation using the Jaccard Index (j) above and the split kmer length (k): (-1/(2k+1))*ln(2j/(1+j)) for 0<j≤1 or 1 for j=0 |
| SNPs | Number of split kmers found in both samples where the middle base is an A, C, G or T but differs between files |
| SNP distance | The ratio of SNPs to matches: SNPs/matches |
| Column | Description |
|---|---|
| File | The name of the split kmer file |
| Cluster__autocolour | An index for the cluster containing the file |
Note: The __autocolour suffix to the Cluster column is to allow automatic colouring when the file is opened in MicroReact
For all clusters comprising two or more samples a samples file will be produced. These files can be used as input for many SKA subcommands.
A graph connecting samples using the cutoffs defined by the user is output and can be visualised along with the clusters.txt file in MicroReact. By default samples are only included in the dot file if they appear in a cluster of two or more samples. To include all samples in the file, use the -S flag.
ska distance [options] <split kmer files>
Options:
-c Do not print clusters files.
-d Do not print distances file.
-h Print this help.
-f <file> File of split kmer file names. These will be added to or
used as an alternative input to the list provided on the
command line.
-i <float> Identity cutoff for defining clusters. Isolates will be
clustered if they share at least this proportion of the
split kmers in the file with fewer kmers and pass the SNP
cutoff. [Default = 0.9]
-o <file> Prefix for output files. [Default = distances]
-s <int> SNP cutoff for defining clusters. Isolates will be clustered
if they are separated by fewer than this number of SNPs and
pass the identity cutoff. [Default = 20]
-S Include singletons in dot file