---
name: "mash-triangle"
section: "1"
platform: "linux"
package: "mash"
version: "2.3+dfsg-6+b2"
release: "Debian 13 (Trixie)"
platforms: ["linux"]
url: "https://new.carta.tech/mash-triangle"
---

# mash-triangle(1)

estimate a lower-triangular distance matrix

## NAME

mash-triangle - estimate a lower-triangular distance matrix

## SYNOPSIS

**mash triangle** \[options\] \<seq1\> \[\<seq2\>\] ...

## DESCRIPTION

Estimate the distance of each input sequence to every other input sequence.  Outputs a lower-triangular distance matrix in relaxed Phylip format. The input sequences can be fasta or fastq, gzipped or not, or Mash sketch files (.msh) with matching k-mer sizes. Input files can also be files of file names (see -l). If more than one input file is provided, whole files are compared by default (see -i).

## OPTIONS

- **-h**

  Help

- **-p** \<int\>

  Parallelism. This many threads will be spawned for processing. \[1\]

### Input

- **-l**

  List input. Each query file contains a list of sequence files, one per line. The reference file is not affected.

### Output

- **-C**

  Use comment fields for sequence names instead of IDs.

- **-E**

  Output edge list instead of Phylip matrix, with fields \[seq1, seq2, dist, p-val, shared-hashes\].

- **-v** \<num\>

  Maximum p-value to report in edge list. Implies -E. (0-1) \[1.0\]

- **-d** \<num\>

  Maximum distance to report in edge list. Implies -E. (0-1) \[1.0\]

### Sketching

- **-k** \<int\>

  K-mer size. Hashes will be based on strings of this many nucleotides. Canonical nucleotides are used by default (see Alphabet options below). (1-32) \[21\]

- **-s** \<int\>

  Sketch size. Each sketch will have at most this many non-redundant min-hashes. \[1000\]

- **-i**

  Sketch individual sequences, rather than whole files, e.g. for multi-fastas of single-chromosome genomes or pair-wise gene comparisons.

- **-w** \<num\>

  Probability threshold for warning about low k-mer size. (0-1) \[0.01\]

- **-r**

  Input is a read set. See Reads options below. Incompatible with **-i**.

### Sketching (reads)

- **-b** \<size\>

  Use a Bloom filter of this size (raw bytes or with K/M/G/T) to filter out unique k-mers. This is useful if exact filtering with **-m** uses too much memory. However, some unique k-mers may pass erroneously, and copies cannot be counted beyond 2. Implies **-r**.

- **-m** \<int\>

  Minimum copies of each k-mer required to pass noise filter for reads. Implies **-r**. \[1\]

- **-c** \<num\>

  Target coverage. Sketching will conclude if this coverage is reached before the end of the input file (estimated by average k-mer multiplicity). Implies **-r**.

- **-g** \<size\>

  Genome size. If specified, will be used for p-value calculation instead of an estimated size from k-mer content. Implies **-r**.

### Sketching (alphabet)

- **-n**

  Preserve strand (by default, strand is ignored by using canonical DNA k-mers, which are alphabetical minima of forward-reverse pairs). Implied if an alphabet is specified with **-a** or **-z**.

- **-a**

  Use amino acid alphabet (A-Z, except BJOUXZ). Implies **-n**, **-k** 9.

- **-z** \<text\>

  Alphabet to base hashes on (case ignored by default; see **-Z**). K-mers with other characters will be ignored. Implies **-n**.

- **-Z**

  Preserve case in k-mers and alphabet (case is ignored by default). Sequence letters whose case is not in the current alphabet will be skipped when sketching.

## SEE ALSO

[mash(1)](/mash/1.md)