emdepth

README

![GoDoc] (https://godoc.org/github.com/brentp/goleft/emdepth?status.png)

emdepth

-- import "github.com/brentp/goleft/emdepth"

emdepth uses a simplified EM algorithm to assign copy-number given a set of depths. it is based off the implementation in cn.mops. Like cn.mops, it works best with more samples and it assumes that most samples will have copy-number 2. emdepth consists of a single function EMDepth that iteratively assigns depths to copy-numbers, adjusts the center of each copy-number bin. and re-assigns... This package does no normalization and therefore expects incoming data to be normalized.

Documentation

Overview

Package emdepth uses a simplified EM algorithm to assign copy-number given a set of depths. it is based off the implementation in cn.mops. Like cn.mops, it works best with more samples and it assumes that most samples will have copy-number 2. emdepth consists of a single function EMDepth that iteratively assigns depths to copy-numbers, adjusts the center of each copy-number bin. and re-assigns... This package does no normalization and therefore expects incoming data to be normalized.

Index

• type CNV
  • func (c *CNV) String() string
• type Cache
  • func (c *Cache) Add(e *EMD) []*CNV
  • func (c *Cache) Clear(p *Position) []*CNV
• type EMD
  • func EMDepth(depths []float32, p Position) *EMD
  • func (e *EMD) CN() []int
  • func (e *EMD) Log2FC() []float64
  • func (e *EMD) Same(o *EMD) (non2 []int, changed []int, pct float64)
  • func (e *EMD) Type(d float32) int
• type Position
  • func (p *Position) String() string

Types

type CNV

type CNV struct {
	SampleI  int
	Depth    []float32
	Position []Position
	Log2FC   []float32
	CN       []int
	PSize    int
}

CNV holds the information for a CNV for a single sample.

func (*CNV) String

func (c *CNV) String() string

type Cache

type Cache struct {
	// contains filtered or unexported fields
}

Cache is a way to track depth states.eps As new items are added to the cache, the value from EMD.Same is compared. Each time an EMD is added, a slice of Regions that have ended is returned.

func (*Cache) Add

func (c *Cache) Add(e *EMD) []*CNV

func (*Cache) Clear

func (c *Cache) Clear(p *Position) []*CNV

type EMD

type EMD struct {
	Lambda   []float64
	Depths   []float32
	Position Position
	// contains filtered or unexported fields
}

EMD holds the posterior maximum depth for each copy-number

func EMDepth

func EMDepth(depths []float32, p Position) *EMD

EMDepth returns a slice of integer copy-numbers (CNs) corresponding to the given normalized depths. It uses a simple EM to assign depths to copy-numbers bins with a preference for CN 2. And to adjust the mean depth of each bin after each iteration. I is a unique identifier for the depths that can be use to assiciate the returned struct with position info.

func (*EMD) CN

func (e *EMD) CN() []int

CN finds the posterior maximum CN for each sample.

func (*EMD) Log2FC

func (e *EMD) Log2FC() []float64

Log2FC is the fold-change relative to copy-number 2.

func (*EMD) Same

func (e *EMD) Same(o *EMD) (non2 []int, changed []int, pct float64)

func (*EMD) Type

func (e *EMD) Type(d float32) int

Type returns the copy-number for the given depth

type Position

type Position struct {
	Start uint32
	End   uint32
}

func (*Position) String

func (p *Position) String() string