org.broadinstitute.hellbender.utils.activityprofile

Class ActivityProfile

java.lang.Object

org.broadinstitute.hellbender.utils.activityprofile.ActivityProfile

Direct Known Subclasses:

BandPassActivityProfile

public class ActivityProfile
extends java.lang.Object

Class holding information about per-base activity scores for assembly region traversal

Field Summary

Fields

Modifier and Type	Field and Description
protected double	activeProbThreshold
protected int	contigLength A cached value of the regionStartLoc contig length, to make calls to getCurrentContigLength efficient
protected int	maxProbPropagationDistance
protected SimpleInterval	regionStartLoc
protected SimpleInterval	regionStopLoc
protected java.util.Set<SimpleInterval>	restrictToIntervals
protected htsjdk.samtools.SAMFileHeader	samHeader
protected java.util.List<ActivityProfileState>	stateList

Constructor Summary

Constructors

Constructor and Description
ActivityProfile(int maxProbPropagationDistance, double activeProbThreshold, htsjdk.samtools.SAMFileHeader header) Create a new empty ActivityProfile
ActivityProfile(int maxProbPropagationDistance, double activeProbThreshold, java.util.Set<SimpleInterval> intervals, htsjdk.samtools.SAMFileHeader header) Create a empty ActivityProfile, restricting output to profiles overlapping intervals, if not null

Method Summary

Modifier and Type	Method and Description
void	add(ActivityProfileState state) Add the next ActivityProfileState to this profile.
java.lang.String	getContig()
int	getEnd()
protected SimpleInterval	getLocForOffset(SimpleInterval relativeLoc, int offset) Helper function that gets the interval for a site offset from relativeLoc, protecting ourselves from falling off the edge of the contig.
int	getMaxProbPropagationDistance() How far away can probability mass be moved around in this profile? This distance puts an upper limit on how far, in bp, we will ever propagate probability mass around when adding a new ActivityProfileState.
protected double[]	getProbabilitiesAsArray() Get the probabilities of the states as a single linear array of doubles
SimpleInterval	getSpan() Get the span of this activity profile, which is from the start of the first state to the stop of the last
protected java.util.List<ActivityProfileState>	getStateList() Get the list of activity profile results in this object
boolean	isEmpty() Is this profile empty? (ie., does it contain no ActivityProfileStates?)
java.util.List<AssemblyRegion>	popReadyAssemblyRegions(int assemblyRegionExtension, int minRegionSize, int maxRegionSize, boolean forceConversion) Get the next completed assembly regions from this profile, and remove all states supporting them from this profile Takes the current profile and finds all of the active / inactive from the start of the profile that are ready.
protected java.util.Collection<ActivityProfileState>	processState(ActivityProfileState justAddedState) Process justAddedState, returning a collection of derived states that actually be added to the stateList The purpose of this function is to transform justAddedStates, if needed, into a series of atomic states that we actually want to track.
int	size() How many profile results are in this profile?
java.lang.String	toString()

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

stateList

protected final java.util.List<ActivityProfileState> stateList

restrictToIntervals

protected final java.util.Set<SimpleInterval> restrictToIntervals

maxProbPropagationDistance

protected final int maxProbPropagationDistance

activeProbThreshold

protected final double activeProbThreshold

regionStartLoc

protected SimpleInterval regionStartLoc

regionStopLoc

protected SimpleInterval regionStopLoc

samHeader

protected htsjdk.samtools.SAMFileHeader samHeader

contigLength

protected int contigLength

A cached value of the regionStartLoc contig length, to make calls to getCurrentContigLength efficient

Constructor Detail

ActivityProfile

public ActivityProfile(int maxProbPropagationDistance,
                       double activeProbThreshold,
                       htsjdk.samtools.SAMFileHeader header)

Create a new empty ActivityProfile

Parameters:

maxProbPropagationDistance - region probability propagation distance beyond its maximum size

activeProbThreshold - threshold for the probability of an activity profile state being active

ActivityProfile

public ActivityProfile(int maxProbPropagationDistance,
                       double activeProbThreshold,
                       java.util.Set<SimpleInterval> intervals,
                       htsjdk.samtools.SAMFileHeader header)

Create a empty ActivityProfile, restricting output to profiles overlapping intervals, if not null

Parameters:

maxProbPropagationDistance - region probability propagation distance beyond its maximum size

activeProbThreshold - threshold for the probability of a profile state being active

intervals - only include states that are within these intervals, if not null

Method Detail

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

getMaxProbPropagationDistance

public int getMaxProbPropagationDistance()

How far away can probability mass be moved around in this profile? This distance puts an upper limit on how far, in bp, we will ever propagate probability mass around when adding a new ActivityProfileState. For example, if the value of this function is 10, and you are looking at a state at bp 5, and we know that no states beyond 5 + 10 will have their probability propagated back to that state.

Returns:

a positive integer distance in bp

size

public int size()

How many profile results are in this profile?

Returns:

the number of profile results

isEmpty

public boolean isEmpty()

Is this profile empty? (ie., does it contain no ActivityProfileStates?)

Returns:

true if the profile is empty (ie., contains no ActivityProfileStates)

getSpan

public SimpleInterval getSpan()

Get the span of this activity profile, which is from the start of the first state to the stop of the last

Returns:

a potentially null SimpleInterval. Will be null if this profile is empty

getContig

public java.lang.String getContig()

getEnd

public int getEnd()

getStateList

protected java.util.List<ActivityProfileState> getStateList()

Get the list of activity profile results in this object

Returns:

a non-null, ordered list of activity profile results

getProbabilitiesAsArray

protected double[] getProbabilitiesAsArray()

Get the probabilities of the states as a single linear array of doubles

Returns:

a non-null array

getLocForOffset

protected SimpleInterval getLocForOffset(SimpleInterval relativeLoc,
                                         int offset)

Helper function that gets the interval for a site offset from relativeLoc, protecting ourselves from falling off the edge of the contig.

Parameters:

relativeLoc - the location offset is relative to

offset - the offset from relativeLoc where we'd like to create a GenomeLoc

Returns:

a SimpleInterval with relativeLoc.start + offset, if this is on the contig, null otherwise

add

public void add(ActivityProfileState state)

Add the next ActivityProfileState to this profile. Must be contiguous with the previously added result, or an IllegalArgumentException will be thrown

Parameters:

state - a well-formed ActivityProfileState result to incorporate into this profile

processState

protected java.util.Collection<ActivityProfileState> processState(ActivityProfileState justAddedState)

Process justAddedState, returning a collection of derived states that actually be added to the stateList The purpose of this function is to transform justAddedStates, if needed, into a series of atomic states that we actually want to track. For example, if state is for soft clips, we transform that single state into a list of states that surround the state up to the distance of the soft clip. Can be overridden by subclasses to transform states in any way There's no particular contract for the output states, except that they can never refer to states beyond the current end of the stateList unless the explicitly include preceding states before the reference. So for example if the current state list is [1, 2, 3] this function could return [1,2,3,4,5] but not [1,2,3,5].

Parameters:

justAddedState - the state our client provided to use to add to the list

Returns:

a list of derived states that should actually be added to this profile's state list

popReadyAssemblyRegions

public java.util.List<AssemblyRegion> popReadyAssemblyRegions(int assemblyRegionExtension,
                                                              int minRegionSize,
                                                              int maxRegionSize,
                                                              boolean forceConversion)

Get the next completed assembly regions from this profile, and remove all states supporting them from this profile Takes the current profile and finds all of the active / inactive from the start of the profile that are ready. By ready we mean unable to have their probability modified any longer by future additions to the profile. The regions that are popped off the profile take their states with them, so the start of this profile will always be after the end of the last region returned here. The regions are returned sorted by genomic position. This function may not return anything in the list, if no regions are ready No returned region will be larger than maxRegionSize.

Parameters:

assemblyRegionExtension - the extension value to provide to the constructed regions

minRegionSize - the minimum region size, in the case where we have to cut up regions that are too large

maxRegionSize - the maximize size of the returned region

forceConversion - if true, we'll return a region whose end isn't sufficiently far from the end of the stateList. Used to close out the active region when we've hit some kind of end (such as the end of the contig)

Returns:

a non-null list of active regions