Python port of the R Bioconductor `seqlogo` package

These details have not been verified by PyPI

Project links

Homepage

GitHub Statistics

View statistics for this project via Libraries.io, or by using our public dataset on Google BigQuery

Project description

seqlogo

Python port of Bioconductor's seqLogo served by WebLogo

Overview

In the field of bioinformatics, a common task is to look for sequence motifs at different sites along the genome or within a protein sequence. One aspect of this analysis involves creating a variant of a Position Matrix (PM): Position Frequency Matrix (PFM), Position Probability Matrix (PPM), and Position Weight Matrix (PWM). The formal format for a PWM file can be found here.

Specification

A PM file can be just a plain text, whitespace delimited matrix, such that the number of columns matches the number of letters in your desired alphabet and the number of rows is the number of positions in your sequence. Any comment lines that start with # will be skipped.

Note: TRANSFAC matrix and MEME Motif formats are not directly supported.

Where is the probability that at position, letter is seen.

This is often generated in a frequentist fashion. If a pipeline tallies all observed letters at each position, this is called a Position Frequency Matrix (PFM).

The PFM can be converted to a PPM in a straight-forward manner, creating a matrix that for any given position and letter, the probability of that letter at that position is reported.

A PWM is the PPM converted into log-likelihood. Pseudocounts can be applied to prevent probabilities of 0 from turing into -inf in the conversion process. Lastly, each position's log-likelihood is corrected for some background probability for every given letter in the selected alphabet.

Features

seqlogo can use any PM as entry points for analysis (from a file or in array formats) and, subsequently, plot the sequence logos.
seqlogo was written to support BIOINF 529 :Bioinformatics Concepts and Algorithms at the University of Michigan in the Department of Computational Medicine & Bioinformatics.
seqlogo attempts to blend the user-friendly api of Bioconductor's seqLogo and the rendering power of the WebLogoPython API.

seqlogo supports the following alphabets:

Alphabet name	Alphabet Letters
`"DNA"`	`"ACGT"`
`"reduced DNA"`	`"ACGTN-"`
`"ambig DNA"`	`"ACGTRYSWKMBDHVN-"`
`"RNA"`	`"ACGU"`
`"reduced RNA"`	`"ACGUN-"`
`"ambig RNA"`	`"ACGURYSWKMBDHVN-"`
`"AA"`	`"ACDEFGHIKLMNPQRSTVWY"`
`"reduced AA"`	`"ACDEFGHIKLMNPQRSTVWYX*-"`
`"ambig AA"`	`"ACDEFGHIKLMNOPQRSTUVWYBJZX*-"`
(Bolded alphabet names are the most commonly used)

seqlogo can also render sequence logos in a number of formats:
- "svg" (default)
- "eps"
- "pdf"
- "jpeg"
- "png"
All plots can be rendered in 4 different sizes:
- "small": 3.54" wide
- "medium": 5" wide
- "large": 7.25" wide
- "xlarge": 10.25" wide

Note: all sizes taken from this publication guide from Science Magazine.

Recommended settings:

For best results, implement seqlogo within a IPython/Jupyter environment (for inline plotting purposes).
Initially written for Python 3.7, but has shown to work in versions 3.5+ (Python 2.7 is not supported)

Setup

Minimal Requirements:

numpy
pandas
weblogo

Note: it is strongly encouraged that jupyter is installed as well.

`conda` environment:

To produce the ideal virtual environment that will run seqlogo on a conda-based build, clone the repo or download the environment.yml within the repo. Then run the following command:

$ conda env create -f environment.yml

Installation

To install using pip:

$ pip install seqlogo

To install using conda

$ conda install -c bioconda seqlogo

Or install from GitHub directly

$ pip install git+https://github.com/betteridiot/seqlogo.git#egg=seqlogo

Quickstart

Importing

import numpy as np
import pandas as pd
import seqlogo

Generate some PM data (without frequency data)

For many demonstrations that speak to PWMs, they are often started with PPM data. Many packages preclude sequence logo generation from this entry point. However, seqlogo can handle it just fine. One point to make though is that if no count data is provided, seqlogo just generates the PFM data by multiplying the probabilities by 100. This is only for weblogolib compatability.

# Setting seed for demonstration purposes
>>> np.random.seed(42)

# Making a fake PPM
>>> random_ppm = np.random.dirichlet(np.ones(4), size=6)
>>> ppm = seqlogo.Ppm(random_ppm)
>>> ppm
          A         C         G         T
0  0.082197  0.527252  0.230641  0.159911
1  0.070375  0.070363  0.024826  0.834435
2  0.161962  0.216972  0.003665  0.617401
3  0.735638  0.098290  0.082638  0.083434
4  0.179898  0.368931  0.280463  0.170708
5  0.498510  0.079138  0.182004  0.240349

Generate some frequency data and convert to PWM

Sometimes the user has frequency data instead of PWM. To construct a Pwm instance that automatically computes Information Content and PWM values, the user can use the seqlogo.pfm2pwm() function.

# Setting seed for demonstration purposes
>>> np.random.seed(42)

# Making some fake Position Frequency Data (PFM)
>>> pfm = pd.DataFrame(np.random.randint(0, 36, size=(8, 4)))

# Convert to Position Weight Matrix (PWM)
>>> pwm = seqlogo.pfm2pwm(pfm)
>>> pwm
          A         C         G         T
0  0.698830 -0.301170 -1.301170  0.213404
1  0.263034  0.552541 -0.584962 -0.584962
2  0.148523  0.754244  0.148523 -3.375039
3  0.182864 -4.209453  0.314109  0.648528
4 -4.000000  0.321928  1.000000 -0.540568
5 -0.222392 -0.029747  0.085730  0.140178
6  0.697437  0.597902 -2.209453 -0.624491
7  0.736966 -0.584962  0.502500 -2.000000

`seqlogo.CompletePm` demo

Here is a quickstart guide on how to leverage the power of seqlogo.CompletePm

# Setting seed for demonstration purposes
>>> np.random.seed(42)

# Making a fake PWM
>>> random_ppm = np.random.dirichlet(np.ones(4), size=6)
>>> cpm = seqlogo.CompletePM(ppm = random_ppm)

# Pfm was imputed
>>> print(cpm.pfm)
    A   C   G   T
0   8  52  23  15
1   7   7   2  83
2  16  21   0  61
3  73   9   8   8
4  17  36  28  17
5  49   7  18  24

# Shows the how the PPM data was formatted
>>> print(cpm.ppm)
          A         C         G         T
0  0.082197  0.527252  0.230641  0.159911
1  0.070375  0.070363  0.024826  0.834435
2  0.161962  0.216972  0.003665  0.617401
3  0.735638  0.098290  0.082638  0.083434
4  0.179898  0.368931  0.280463  0.170708
5  0.498510  0.079138  0.182004  0.240349

# Computing the PWM using default background and pseudocounts
>>> print(cpm.pwm)
          A         C         G         T
0 -1.604773  1.076564 -0.116281 -0.644662
1 -1.828788 -1.829031 -3.331983  1.738871
2 -0.626276 -0.204418 -6.091862  1.304279
3  1.557068 -1.346815 -1.597049 -1.583223
4 -0.474749  0.561423  0.165882 -0.550396
5  0.995695 -1.659494 -0.457960 -0.056800

# See the consensus sequence
>>> print(cpm.consensus)
CTTACA

# See the Information Content
>>> print(cpm.ic)
0    0.305806
1    1.110856
2    0.637149
3    0.748989
4    0.074286
5    0.268034
dtype: float64

Plot the sequence logo with information content scaling

# Setting seed for demonstration purposes
>>> np.random.seed(42)

# Making a fake PWM
>>> random_ppm = np.random.dirichlet(np.ones(4), size=6)
>>> ppm = seqlogo.Ppm(random_ppm)
>>> seqlogo.seqlogo(ppm, ic_scale = False, format = 'svg', size = 'medium')

The above code will produce:

Plot the sequence logo with no information content scaling

# Setting seed for demonstration purposes
>>> np.random.seed(42)

# Making a fake PWM
>>> random_ppm = np.random.dirichlet(np.ones(4), size=6)
>>> ppm = seqlogo.Ppm(random_ppm)
>>> seqlogo.seqlogo(ppm, ic_scale = False, format = 'svg', size = 'medium')

The above code will produce:

Documentation

seqlogo exposes 5 classes to the user for handling PM data:

seqlogo.Pm: the base class for all other specialized PM subclasses
seqlogo.Pfm: The class used for handling PFM data
seqlogo.Ppm: The class used for handling PPM data
seqlogo.Pwm: The class used for handling PWM data
seqlogo.CompletePm: This final class will take any/all of the other PM subclass data and compute any of the other missing data. That is, if the user only provides a seqlogo.Pfm and passes it to seqlogo.CompletePm, it will solve for the PPM, PWM, consensus sequence, and information content.

Additionally, seqlogo also provides 6 methods for converting PM structures:

seqlogo.pfm2ppm: converts a PFM to a PPM
seqlogo.pfm2pwm: converts a PFM to a PWM
seqlogo.ppm2pfm: converts a PPM to a PFM
seqlogo.ppm2pwm: converts a PPM to a PWM
seqlogo.pwm2pfm: converts a PWM to a PFM
seqlogo.pwm2ppm: converts a PWM to a PPM

The signatures for each item above are as follows:

Classes

seqlogo.CompletePm(pfm = None, ppm = None, pwm = None, background = None, pseudocount = None,
                 alphabet_type = 'DNA', alphabet = None, default_pm = 'ppm'):
    """
    Creates the CompletePm instance. If the user does not define any `pm_filename_or_array`,
    it will be initialized to empty. Will generate all other attributes as soon
    as a `pm_filename_or_array` is supplied.

    Args:
        pfm (str or `numpy.ndarray` or `pandas.DataFrame` or Pm): The user supplied
            PFM. If it is a filename, the file will be opened
            and parsed. If it is an `numpy.ndarray` or `pandas.DataFrame`,
            it will just be assigned. (default: None, skips '#' comment lines)
        ppm (str or `numpy.ndarray` or `pandas.DataFrame` or Pm): The user supplied
            PPM. If it is a filename, the file will be opened
            and parsed. If it is an `numpy.ndarray` or `pandas.DataFrame`,
            it will just be assigned. (default: None, skips '#' comment lines)
        pwm (str or `numpy.ndarray` or `pandas.DataFrame` or Pm): The user supplied
            PWM. If it is a filename, the file will be opened
            and parsed. If it is an `numpy.ndarray` or `pandas.DataFrame`,
            it will just be assigned. (default: None, skips '#' comment lines)
        background (constant or Collection): Offsets used to calculate background letter probabilities (defaults: If 
            using an Nucleic Acid alphabet: 0.25; if using an Aminio Acid alphabet: Robinson-Robinson Frequencies)
        pseudocount (constant): Some constant to offset PPM conversion to PWM to prevent -/+ inf. (defaults to 1e-10)
        alphabet_type (str): Desired alphabet to use. Order matters (default: 'DNA')
            "DNA" := "ACGT"
            "reduced DNA" := "ACGTN-"
            "ambig DNA" := "ACGTRYSWKMBDHVN-"
            "RNA" := "ACGU"
            "reduced RNA" := "ACGUN-"
            "ambig RNA" := "ACGURYSWKMBDHVN-"
            "AA" : = "ACDEFGHIKLMNPQRSTVWY"
            "reduced AA" := "ACDEFGHIKLMNPQRSTVWYX*-"
            "ambig AA" := "ACDEFGHIKLMNOPQRSTUVWYBJZX*-"
            "custom" := None
            (default: 'DNA')
        alphabet (str): if 'custom' is selected or a specialize alphabet is desired, this accepts a string (default: None)
        default_pm (str): which of the 3 pm's do you want to call '*home*'? (default: 'ppm')
    """

seqlogo.Pm(pm_filename_or_array = None, pm_type = 'ppm', alphabet_type = 'DNA', alphabet = None, 
    background = None, pseudocount = None):
    """Initializes the Pm

    Creates the Pm instance. If the user does not define `pm_filename_or_array`,
    it will be initialized to empty. Will generate all other attributes as soon
    as a `pm_filename_or_array` is supplied.

    Args:
        pm_filename_or_array (str or `numpy.ndarray` or `pandas.DataFrame` or Pm): The user supplied
            PM. If it is a filename, the file will be opened
            and parsed. If it is an `numpy.ndarray` or `pandas.DataFrame`,
            it will just be assigned. (default: None, skips '#' comment lines)
        alphabet_type (str): Desired alphabet to use. Order matters (default: 'DNA')
            "DNA" := "ACGT"
            "reduced DNA" := "ACGTN-"
            "ambig DNA" := "ACGTRYSWKMBDHVN-"
            "RNA" := "ACGU"
            "reduced RNA" := "ACGUN-"
            "ambig RNA" := "ACGURYSWKMBDHVN-"
            "AA" : = "ACDEFGHIKLMNPQRSTVWY"
            "reduced AA" := "ACDEFGHIKLMNPQRSTVWYX*-"
            "ambig AA" := "ACDEFGHIKLMNOPQRSTUVWYBJZX*-"
            "custom" := None
            (default: 'DNA')
        alphabet (str): if 'custom' is selected or a specialize alphabet is desired, this accepts a string (default: None)
        background (constant or Collection): Offsets used to calculate background letter probabilities (defaults: If 
            using an Nucleic Acid alphabet: 0.25; if using an Aminio Acid alphabet: Robinson-Robinson Frequencies)
        pseudocount (constant): Some constant to offset PPM conversion to PWM to prevent -/+ inf. (default: 1e-10)
    """

seqlogo.Pfm(pfm_filename_or_array = None, pm_type = 'pfm', alphabet_type = 'DNA', alphabet = None, 
    background = None, pseudocount = None):
    """Initializes the Pfm

    Creates the Pfm instance. If the user does not define `pfm_filename_or_array`,
    it will be initialized to empty. Will generate all other attributes as soon
    as a `pfm_filename_or_array` is supplied.

    Args:
        pfm_filename_or_array (str or `numpy.ndarray` or `pandas.DataFrame` or Pm): The user supplied
            PFM. If it is a filename, the file will be opened
            and parsed. If it is an `numpy.ndarray` or `pandas.DataFrame`,
            it will just be assigned. (default: None, skips '#' comment lines)
        alphabet_type (str): Desired alphabet to use. Order matters (default: 'DNA')
            "DNA" := "ACGT"
            "reduced DNA" := "ACGTN-"
            "ambig DNA" := "ACGTRYSWKMBDHVN-"
            "RNA" := "ACGU"
            "reduced RNA" := "ACGUN-"
            "ambig RNA" := "ACGURYSWKMBDHVN-"
            "AA" : = "ACDEFGHIKLMNPQRSTVWY"
            "reduced AA" := "ACDEFGHIKLMNPQRSTVWYX*-"
            "ambig AA" := "ACDEFGHIKLMNOPQRSTUVWYBJZX*-"
            "custom" := None
            (default: 'DNA')
        alphabet (str): if 'custom' is selected or a specialize alphabet is desired, this accepts a string (default: None)
        background (constant or Collection): Offsets used to calculate background letter probabilities (defaults: If 
            using an Nucleic Acid alphabet: 0.25; if using an Aminio Acid alphabet: Robinson-Robinson Frequencies)
        pseudocount (constant): Some constant to offset PPM conversion to PWM to prevent -/+ inf. (default: 1e-10)
    """

seqlogo.Ppm(ppm_filename_or_array = None, pm_type = 'ppm', alphabet_type = 'DNA', alphabet = None, 
    background = None, pseudocount = None):
    """Initializes the Ppm

    Creates the Ppm instance. If the user does not define `ppm_filename_or_array`,
    it will be initialized to empty. Will generate all other attributes as soon
    as a `ppm_filename_or_array` is supplied.

    Args:
        ppm_filename_or_array (str or `numpy.ndarray` or `pandas.DataFrame` or Pm): The user supplied
            PPM. If it is a filename, the file will be opened
            and parsed. If it is an `numpy.ndarray` or `pandas.DataFrame`,
            it will just be assigned. (default: None, skips '#' comment lines)
        alphabet_type (str): Desired alphabet to use. Order matters (default: 'DNA')
            "DNA" := "ACGT"
            "reduced DNA" := "ACGTN-"
            "ambig DNA" := "ACGTRYSWKMBDHVN-"
            "RNA" := "ACGU"
            "reduced RNA" := "ACGUN-"
            "ambig RNA" := "ACGURYSWKMBDHVN-"
            "AA" : = "ACDEFGHIKLMNPQRSTVWY"
            "reduced AA" := "ACDEFGHIKLMNPQRSTVWYX*-"
            "ambig AA" := "ACDEFGHIKLMNOPQRSTUVWYBJZX*-"
            "custom" := None
            (default: 'DNA')
        alphabet (str): if 'custom' is selected or a specialize alphabet is desired, this accepts a string (default: None)
        background (constant or Collection): Offsets used to calculate background letter probabilities (defaults: If 
            using an Nucleic Acid alphabet: 0.25; if using an Aminio Acid alphabet: Robinson-Robinson Frequencies)
        pseudocount (constant): Some constant to offset PPM conversion to PWM to prevent -/+ inf. (default: 1e-10)
   """
   
seqlogo.Pwm(pwm_filename_or_array = None, pm_type = 'pwm', alphabet_type = 'DNA', alphabet = None, 
    background = None, pseudocount = None):
    """Initializes the Pwm

    Creates the Pwm instance. If the user does not define `pwm_filename_or_array`,
    it will be initialized to empty. Will generate all other attributes as soon
    as a `pwm_filename_or_array` is supplied.

    Args:
        pwm_filename_or_array (str or `numpy.ndarray` or `pandas.DataFrame` or Pm): The user supplied
            PWM. If it is a filename, the file will be opened
            and parsed. If it is an `numpy.ndarray` or `pandas.DataFrame`,
            it will just be assigned. (default: None, skips '#' comment lines)
        alphabet_type (str): Desired alphabet to use. Order matters (default: 'DNA')
            "DNA" := "ACGT"
            "reduced DNA" := "ACGTN-"
            "ambig DNA" := "ACGTRYSWKMBDHVN-"
            "RNA" := "ACGU"
            "reduced RNA" := "ACGUN-"
            "ambig RNA" := "ACGURYSWKMBDHVN-"
            "AA" : = "ACDEFGHIKLMNPQRSTVWY"
            "reduced AA" := "ACDEFGHIKLMNPQRSTVWYX*-"
            "ambig AA" := "ACDEFGHIKLMNOPQRSTUVWYBJZX*-"
            "custom" := None
            (default: 'DNA')
        alphabet (str): if 'custom' is selected or a specialize alphabet is desired, this accepts a string (default: None)
        background (constant or Collection): Offsets used to calculate background letter probabilities (defaults: If 
            using an Nucleic Acid alphabet: 0.25; if using an Aminio Acid alphabet: Robinson-Robinson Frequencies)
        pseudocount (constant): Some constant to offset PPM conversion to PWM to prevent -/+ inf. (default: 1e-10)
   """

Conversion Methods

seqlogo.pfm2ppm(pfm):
    """Converts a Pfm to a ppm array

    Args:
        pfm (Pfm): a fully initialized Pfm

    Returns:
        (np.array): converted values
    """
    
seqlogo.pfm2pwm(pfm, background = None, pseudocount = None):
    """Converts a Pfm to a pwm array

    Args:
        pfm (Pfm): a fully initialized Pfm
        background: accounts for relative weights from background. Must be a constant or same number of columns as Pwm (default: None)
        pseudocount (const): The number used to offset log-likelihood conversion from probabilites (default: None -> 1e-10)

    Returns:
        (np.array): converted values
    """

seqlogo.ppm2pfm(ppm):
    """Converts a Ppm to a pfm array

    Args:
        ppm (Ppm): a fully initialized Ppm

    Returns:
        (np.array): converted values
    """

seqlogo.ppm2pwm(ppm, background= None, pseudocount = None):
    """Converts a Ppm to a pwm array

    Args:
        ppm (Ppm): a fully initialized Ppm
        background: accounts for relative weights from background. Must be a constant or same number of columns as Pwm (default: None)
        pseudocount (const): The number used to offset log-likelihood conversion from probabilites (default: None -> 1e-10)

    Returns:
        (np.array): converted values

    Raises:
        ValueError: if the pseudocount isn't a constant or the same length as sequence
    """

seqlogo.pwm2pfm(pwm, background = None, pseudocount = None):
    """Converts a Pwm to a pfm array

    Args:
        pwm (Pwm): a fully initialized Pwm
        background: accounts for relative weights from background. Must be a constant or same number of columns as Pwm (default: None)
        pseudocount (const): The number used to offset log-likelihood conversion from probabilites (default: None -> 1e-10)

    Returns:
        (np.array): converted values
    """

seqlogo.pwm2ppm(pwm, background = None, pseudocount = None):
    """Converts a Pwm to a ppm array

    Args:
        pwm (Pwm): a fully initialized Pwm
        background: accounts for relative weights from background. Must be a constant or same number of columns as Pwm (default: None)
        pseudocount (const): The number used to offset log-likelihood conversion from probabilites (default: None -> 1e-10)

    Returns:
        (np.array): converted values

    Raises:
        ValueError: if the pseudocount isn't a constant or the same length as sequence
    """

Contributing

Please see our contribution guidelines here

Acknowledgments

Bembom O (2018). seqlogo: Sequence logos for DNA sequence alignments. R package version 1.48.0.
Crooks GE, Hon G, Chandonia JM, Brenner SE WebLogo: A sequence logo generator, Genome Research, 14:1188-1190, (2004).

Project details

These details have not been verified by PyPI

Project links

Homepage

GitHub Statistics

View statistics for this project via Libraries.io, or by using our public dataset on Google BigQuery

Release history Release notifications | RSS feed

5.29.8

Jun 23, 2020

5.29.7

Nov 12, 2019

5.29.6

Aug 14, 2019

5.29.5

Aug 13, 2019

5.29.4

May 9, 2019

5.29.3

Mar 19, 2019

5.29.2

Mar 19, 2019

5.29.1

Feb 1, 2019

5.2.9001

Feb 1, 2019

5.2.10

Jan 23, 2019

This version

5.2.9

Jan 19, 2019

5.2.0

Jan 16, 2019

0.2.4

Jan 16, 2019

0.2.3

Jan 16, 2019

0.2.2

Jan 16, 2019

0.2.1

Jan 16, 2019

0.2.0

Jan 11, 2019

0.1.13

Jan 1, 2019

0.1.12

Dec 28, 2018

0.1.11

Dec 28, 2018

0.1.10

Dec 28, 2018

0.1.9

Dec 28, 2018

0.1.8

Dec 28, 2018

0.1.7

Dec 28, 2018

0.1.6

Dec 28, 2018

0.1.5

Dec 28, 2018

0.1.4

Dec 28, 2018

0.1.3

Dec 28, 2018

0.1.2

Dec 28, 2018

0.1.1

Dec 27, 2018

0.1.0

Dec 27, 2018

0.0.6

Dec 23, 2018

0.0.5

Dec 23, 2018

0.0.4

Dec 23, 2018

0.0.3

Dec 22, 2018

0.0.2

Dec 22, 2018

0.0.1

Dec 22, 2018

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

seqlogo-5.2.9.tar.gz (27.4 kB view hashes)

Uploaded Jan 19, 2019 Source

Hashes for seqlogo-5.2.9.tar.gz

Hashes for seqlogo-5.2.9.tar.gz
Algorithm	Hash digest
SHA256	`4ef1e39ddee3285c354b5349a326e14b5293d77933175dd4c858a34de62f9738`
MD5	`3d87c65f568cb85fc7da2be7b4a4992a`
BLAKE2b-256	`77d56865f19a2c578d73f75eb92b55ee8fd1d52ebda48f43b89c77fa1a0a0372`

seqlogo 5.2.9

Navigation

Verified details

Maintainers

Unverified details

Project links

GitHub Statistics

Meta

Classifiers

Project description

seqlogo

Overview

Specification

Features

Recommended settings:

Setup

Minimal Requirements:

conda environment:

Installation

Quickstart

Importing

Generate some PM data (without frequency data)

Generate some frequency data and convert to PWM

seqlogo.CompletePm demo

Plot the sequence logo with information content scaling

Plot the sequence logo with no information content scaling

Documentation

Classes

Conversion Methods

Contributing

Acknowledgments

Project details

Verified details

Maintainers

Unverified details

Project links

GitHub Statistics

Meta

Classifiers

Release history Release notifications | RSS feed

Download files

Source Distribution

`conda` environment:

`seqlogo.CompletePm` demo