riptide 2.1.1

Reaction Inclusion by Parsimony and Transcript Distribution (RIPTiDe)

RIPTiDe

Reaction Inclusion by Parsimony and Transcript Distribution

---

Transcriptomic analyses of bacteria have become instrumental to our understanding of their responses to changes in their environment. While traditional analyses have been informative, leveraging these datasets within genome-scale metabolic network reconstructions can provide greatly improved context for shifts in pathway utilization and downstream/upstream ramifications for changes in metabolic regulation. Previous techniques for transcript integration have focused on creating maximum consensus with the input datasets. However, these approaches have collectively performed poorly for metabolic predictions even compared to transcript-agnostic approaches of flux minimization that identifies the most efficient patterns of metabolism given certain growth constraints. Our new method, RIPTiDe, combines these concepts and utilizes overall minimization of flux in conjunction with transcriptomic analysis to identify the most energy efficient pathways to achieve growth that include more highly transcribed enzymes. RIPTiDe requires a low level of manual intervention which leads to reduced bias in predictions.

Please cite when using:

Jenior ML, Moutinho TJ, and Papin JA. (2019). Parsimonious transcript data integration improves context-specific predictions of bacterial metabolism in complex environments. bioRxiv 637124; doi: https://doi.org/10.1101/637124

Utilizes python implementation of the gapsplit flux sampler. Please also cite:

Keaty TC and Jensen PA (2019). gapsplit: Efficient random sampling for non-convex constraint-based models. bioRxiv 652917; doi: https://doi.org/10.1101/652917

Dependencies

>=python-3.6.4
>=cobra-0.15.3
>=pandas-0.24.1
>=symengine-0.4.0

Installation

Installation is:

$ pip install riptide

Or from github:

$ pip install git+https://github.com/mjenior/riptide

Usage

from riptide import *

my_model = cobra.io.read_sbml_model('examples/genre.sbml')

transcript_abundances_1 = riptide.read_transcription_file(read_abundances_file='examples/transcriptome1.tsv')
transcript_abundances_2 = riptide.read_transcription_file(read_abundances_file='examples/transcriptome2.tsv')

transcript_abundances_1 = riptide.read_transcription_file('examples/transcriptome1.tsv')
transcript_abundances_2 = riptide.read_transcription_file('examples/transcriptome2.tsv', replicates=True)

riptide_object_1_a = riptide.contextualize(model=my_model, transcription=transcript_abundances_1)
riptide_object_1_b = riptide.contextualize(model=my_model, transcription=transcript_abundances_1, defined='examples/defined.tsv')
riptide_object_2 = riptide.contextualize(model=my_model, transcription=transcript_abundances_2)

Additional parameters for main RIPTiDe functions:

riptide.read_transcription_file()

read_abundances_file : string
    User-provided file name which contains gene IDs and associated transcription values
header : boolean
    Defines if read abundance file has a header that needs to be ignored
    default is no header
replicates : boolean
    Defines if read abundances contains replicates and medians require calculation
    default is no replicates (False)
sep: string
    Defines what character separates entries on each line
    defaults to tab (.tsv)

riptide.contextualize()

model : cobra.Model
    The model to be contextualized (REQUIRED)
transcription : dictionary
    Dictionary of transcript abundances, output of read_transcription_file (REQUIRED)
defined : False or File
    Text file containing reactions IDs for forced inclusion listed on the first line and exclusion 
    listed on the second line (both .csv and .tsv formats supported)
samples : int 
    Number of flux samples to collect, default is 500
percentiles : list of floats
    Percentile cutoffs of transcript abundance for linear coefficient assignments to associated reactions
    Default is [50.0, 62.5, 75.0, 87.5]
coefficients : list of floats
    Linear coefficients to weight reactions based on distribution placement
    Default is [1.0, 0.5, 0.1, 0.01, 0.001]
fraction : float
    Minimum percent of optimal objective value during FBA steps
    Default is 0.75
conservative : bool
    Conservatively remove inactive reactions based on genes
    Default is False
bound : bool
    Bounds each reaction based on transcriptomic constraints
    Default is False
objective : bool
    Sets previous objective function as a constraint with minimum flux equal to user input fraction
    Default is True
set_bounds : bool
    Uses flax variability analysis results from constrained model to set new bounds for all equations
    Default is True

Example stdout report:

Initializing model and integrating transcriptomic data...
Pruning zero flux subnetworks...
Analyzing context-specific flux distributions...

Reactions pruned to 285 from 1129 (74.76% change)
Metabolites pruned to 285 from 1132 (74.82% change)
Flux through the objective DECREASED to ~54.71 from ~65.43 (16.38% change)

RIPTiDe completed in 35 seconds

Resulting RIPTiDe object (class) properties:

• model - Contextualized genome-scale metabolic network reconstruction
• transcriptome - Dictionary of transcriptomic abundances provded by user
• coefficients - Dictionary of linear coefficients assigned to each reaction based on transcript values
• flux_samples - Flux sampling pandas object from constrained model
• flux_variability - Flux variability analysis pandas object from constrained model
• quantile_range - Percentile intervals by which to parse transcript abundance distribution
• linear_coefficient_range - Linear coeeficients assigned to corresponding quantile
• fraction_of_optimum - Minimum specified percentage of optimal objective flux during contextualization

Additional Information

Thank you for your interest in RIPTiDe, for additional questions please email [e-mail].

If you encounter any problems, please file an issue along with a detailed description.

Distributed under the terms of the MIT license, "riptide" is free and open source software