io.github.bigbio.pgatk.io.utils.msnumpress

Class MSNumpress

java.lang.Object

io.github.bigbio.pgatk.io.utils.msnumpress.MSNumpress

public class MSNumpress
extends Object

Field Summary

Fields

Modifier and Type	Field and Description
static String	ACC_NUMPRESS_LINEAR
static String	ACC_NUMPRESS_PIC
static String	ACC_NUMPRESS_SLOF

Constructor Summary

Constructors

Constructor and Description
MSNumpress()

Method Summary

Modifier and Type	Method and Description
static double[]	decode(String cvAccession, byte[] data, int dataSize) Convenience function for decoding binary data encoded by MSNumpress.
static double	decodeFixedPoint(byte[] data)
static int	decodeLinear(byte[] data, int dataSize, double[] result) Decodes data encoded by encodeLinear.
static int	decodePic(byte[] data, int dataSize, double[] result) Decodes data encoded by encodePic
static int	decodeSlof(byte[] data, int dataSize, double[] result) Decodes data encoded by encodeSlof
static void	encodeFixedPoint(double fixedPoint, byte[] result)
protected static int	encodeInt(long x, byte[] res, int resOffset) This encoding works on a 4 byte integer, by truncating initial zeros or ones.
static int	encodeLinear(double[] data, int dataSize, byte[] result, double fixedPoint) Encodes the doubles in data by first using a - lossy conversion to a 4 byte 5 decimal fixed point repressentation - storing the residuals from a linear prediction after first to values - encoding by encodeInt (see above)
static int	encodePic(double[] data, int dataSize, byte[] result) Encodes ion counts by simply rounding to the nearest 4 byte integer, and compressing each integer with encodeInt.
static int	encodeSlof(double[] data, int dataSize, byte[] result, double fixedPoint) Encodes ion counts by taking the natural logarithm, and storing a fixed point representation of this.
static double	optimalLinearFixedPoint(double[] data, int dataSize)
static double	optimalSlofFixedPoint(double[] data, int dataSize)

Methods inherited from class java.lang.Object

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

Field Detail

ACC_NUMPRESS_LINEAR

public static final String ACC_NUMPRESS_LINEAR

See Also:

Constant Field Values

ACC_NUMPRESS_PIC

public static final String ACC_NUMPRESS_PIC

See Also:

Constant Field Values

ACC_NUMPRESS_SLOF

public static final String ACC_NUMPRESS_SLOF

See Also:

Constant Field Values

Constructor Detail

MSNumpress

public MSNumpress()

Method Detail

decode

public static double[] decode(String cvAccession,
                              byte[] data,
                              int dataSize)

Convenience function for decoding binary data encoded by MSNumpress. If the passed cvAccession is one of

ACC_NUMPRESS_LINEAR = "MS:1002312" ACC_NUMPRESS_PIC = "MS:1002313" ACC_NUMPRESS_SLOF = "MS:1002314"

the corresponding decode function will be called.

Parameters:

cvAccession - The PSI-MS obo CV accession of the encoded data.

data - array of double to be encoded

dataSize - number of doubles from data to encode

Returns:

The decoded doubles

encodeInt

protected static int encodeInt(long x,
                               byte[] res,
                               int resOffset)

This encoding works on a 4 byte integer, by truncating initial zeros or ones. If the initial (most significant) half byte is 0x0 or 0xf, the number of such halfbytes starting from the most significant is stored in a halfbyte. This initial count is then followed by the rest of the ints halfbytes, in little-endian order. A count halfbyte c of

0 <= c <= 8 is interpreted as an initial c 0x0 halfbytes 9 <= c <= 15 is interpreted as an initial (c-8) 0xf halfbytes

Ex: int c rest 0 => 0x8 -1 => 0xf 0xf 23 => 0x6 0x7 0x1

Parameters:

x - the int to be encoded

res - the byte array were halfbytes are stored

resOffset - position in res were halfbytes are written

Returns:

the number of resulting halfbytes

encodeFixedPoint

public static void encodeFixedPoint(double fixedPoint,
                                    byte[] result)

decodeFixedPoint

public static double decodeFixedPoint(byte[] data)

optimalLinearFixedPoint

public static double optimalLinearFixedPoint(double[] data,
                                             int dataSize)

encodeLinear

public static int encodeLinear(double[] data,
                               int dataSize,
                               byte[] result,
                               double fixedPoint)

Encodes the doubles in data by first using a - lossy conversion to a 4 byte 5 decimal fixed point repressentation - storing the residuals from a linear prediction after first to values - encoding by encodeInt (see above)

The resulting binary is maximally 8 + dataSize * 5 bytes, but much less if the data is reasonably smooth on the first order.

This encoding is suitable for typical m/z or retention time binary arrays. On a test set, the encoding was empirically show to be accurate to at least 0.002 ppm.

Parameters:

data - array of doubles to be encoded

dataSize - number of doubles from data to encode

result - array were resulting bytes should be stored

fixedPoint - the scaling factor used for getting the fixed point repr. This is stored in the binary and automatically extracted on decoding.

Returns:

the number of encoded bytes

decodeLinear

public static int decodeLinear(byte[] data,
                               int dataSize,
                               double[] result)

Decodes data encoded by encodeLinear.

result vector guaranteed to be shorter or equal to (|data| - 8) * 2

Note that this method may throw a ArrayIndexOutOfBoundsException if it deems the input data to be corrupt, i.e. that the last encoded int does not use the last byte in the data. In addition the last encoded int need to use either the last halfbyte, or the second last followed by a 0x0 halfbyte.

Parameters:

data - array of bytes to be decoded

dataSize - number of bytes from data to decode

result - array were resulting doubles should be stored

Returns:

the number of decoded doubles, or -1 if dataSize < 4 or 4 < dataSize < 8

encodePic

public static int encodePic(double[] data,
                            int dataSize,
                            byte[] result)

Encodes ion counts by simply rounding to the nearest 4 byte integer, and compressing each integer with encodeInt.

The handleable range is therefore 0 -> 4294967294. The resulting binary is maximally dataSize * 5 bytes, but much less if the data is close to 0 on average.

Parameters:

data - array of doubles to be encoded

dataSize - number of doubles from data to encode

result - array were resulting bytes should be stored

Returns:

the number of encoded bytes

decodePic

public static int decodePic(byte[] data,
                            int dataSize,
                            double[] result)

Decodes data encoded by encodePic

result vector guaranteed to be shorter of equal to |data| * 2

Note that this method may throw a ArrayIndexOutOfBoundsException if it deems the input data to be corrupt, i.e. that the last encoded int does not use the last byte in the data. In addition the last encoded int need to use either the last halfbyte, or the second last followed by a 0x0 halfbyte.

Parameters:

data - array of bytes to be decoded (need memorycont. repr.)

dataSize - number of bytes from data to decode

result - array were resulting doubles should be stored

Returns:

the number of decoded doubles

optimalSlofFixedPoint

public static double optimalSlofFixedPoint(double[] data,
                                           int dataSize)

encodeSlof

public static int encodeSlof(double[] data,
                             int dataSize,
                             byte[] result,
                             double fixedPoint)

Encodes ion counts by taking the natural logarithm, and storing a fixed point representation of this. This is calculated as

unsigned short fp = log(d+1) * fixedPoint + 0.5

the result vector is exactly |data| * 2 + 8 bytes long

Parameters:

data - array of doubles to be encoded

dataSize - number of doubles from data to encode

result - array were resulting bytes should be stored

fixedPoint - the scaling factor used for getting the fixed point repr. This is stored in the binary and automatically extracted on decoding.

Returns:

the number of encoded bytes

decodeSlof

public static int decodeSlof(byte[] data,
                             int dataSize,
                             double[] result)

Decodes data encoded by encodeSlof

The result vector will be exactly (|data| - 8) / 2 doubles. returns the number of doubles read, or -1 is there is a problem decoding.

Parameters:

data - array of bytes to be decoded (need memorycont. repr.)

dataSize - number of bytes from data to decode

result - array were resulting doubles should be stored

Returns:

the number of decoded doubles