3.1.1.2. h1_hesc.kb_gen package¶

3.1.1.2.1. Submodules¶

3.1.1.2.2. h1_hesc.kb_gen.chromosomes module¶

Reconstruct the sequences of H1-hESC chromosomes

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Author: Jonathan Karr <jonrkarr@gmail.com>
Date: 2017-07-13
Copyright: 2017, Karr Lab
License: MIT

class h1_hesc.kb_gen.chromosomes.ChromosomesGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Create chromosomes for the knowledge base

clean_and_validate_options()[source]¶

Apply default options and validate options

Options:

data_path(str): path to genome sequence fasta file

gen_components()[source]¶: Construct chromosomes for the knowledge base

get_data()[source]¶: Get data for knowledge base

process_data()[source]¶: Process data for knowledge base_counts

3.1.1.2.3. h1_hesc.kb_gen.compartments module¶

Reconstruct the compartments of a cell

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2018-10-04
Copyright: 2018, Karr Lab
License: MIT

class h1_hesc.kb_gen.compartments.CompartmentsGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Creates compartments and their volumetric fractions for the knowledge base from input files.

Options: * data_path (str): path of manually curated data volumetric

fractions of compartments

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct compartments for the knowledge base

get_data()[source]¶: Get compartment data from input file

3.1.1.2.4. h1_hesc.kb_gen.complexes module¶

Reconstruct protein complexes :Author: Yin Hoon Chew <yinhoon.chew@mssm.edu> :Date: 2017-11-28 :Copyright: 2017, Karr Lab :License: MIT

class h1_hesc.kb_gen.complexes.ComplexesGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Create macromolecular complexes for the knowledge base

Data of protein complexes from Human Recon 2.2, HumanCyc, Corum, and Complex portal are merged into a consensus set. Each recconstructed protein complex contains information of the protein and cofactor subunit as well as their stoichiometric information wherever available.

Each database has different degrees of information. Complex Portal has both stoichiometric and isomer information (e.g. P14618-1, P14618-2) Both Recon and Corum do not have stoichiometric information, but Corum includes protein isomer information (e.g. P78381-1, P78381-2). HumanCyc has stoichiometric information.

A consensus set is compiled by consecutively evaluating and adding each complex entry from each database:

All complex entries in Complex Portal are added. If two complexes have protein subunit(s) that only differ(s) in their isomer information, they are considered the same complex because the mapping of uniprot isomer to ensembl protein is not complete.

Entries from Recon are added if no complex with similar protein subunits could be found in the compiled set

Entries from Corum are added if no complex with similar protein subunits could be found in the compiled set. Isomer information are not considered.

For HumanCyc, because it has an inconsistent definition of complexes compared to other databases, i.e. some HumanCyc complexes are only subsets of complexes from other databases (see example of P03886). we only use its stoichiometric information to fill in the missing stoichiometries of entries in the compiled set where all protein subunits are similar.

Finally, for each entry in the consensus set, if the protein subunits of an entry is the subset of the protein subunits in another entry, we lift the stoichiometry of the superset entry to the subset entry

Options:

id_conversion_path (str): path of gene-transcript-protein id conversion table
manually_curated_data_path (str): path of manually curated enzyme complexes
humancyc_enzyme_path (str): path of HumanCyc data on enzymes
humancyc_transporter_path (str): path of HumanCyc data on transporters
humancyc_geneproduct_path (str): path of HumanCyc data on gene products
humancyc_gene_path (str): path of HumanCyc data on genes
non_protein_conversion (dict): dictionary where the keys are the
non-protein subunit ids used in Complex Portal while the values are references to the species type objects the ids refer to
test_protein (list): list of protein Uniprot IDs for which complexes
are to be retrieved from Corum (for unit testing purpose)
test_complex (list, optional): list of complexes to be reconstructed
(for unit testing purpose)

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct complex species types for the knowledge base

get_data()[source]¶: Get protein complex data from Human Recon 2.2, HumanCyc, Corum, and Complex portal

process_data()[source]¶: Merge data of protein complexes from the various databases

3.1.1.2.5. h1_hesc.kb_gen.core module¶

Generator for the knowledge base

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Author: Jonathan Karr <jonrkarr@gmail.com>
Date: 2018-01-29
Copyright: 2018, Karr Lab
License: MIT

class h1_hesc.kb_gen.core.KbGenerator(component_generators=None, options=None)[source]¶

Bases: wc_kb_gen.core.KbGenerator

Generator for the knowledge base

Options: * id * name * version * component

TaxonGenerator

CompartmentsGenerator

ChromosomesGenerator

GenesTranscriptsExonsProteinsGenerator

RegulatoryModulesGenerator

MetabolitesGenerator

PropertiesGenerator

TranscriptsGenerator

MetabolicNetworkGenerator

ProteinsGenerator

ComplexesGenerator

MetabolicReactionKineticsGenerator

ProteinModificationsGenerator

DEFAULT_COMPONENT_GENERATORS = (<class 'h1_hesc.kb_gen.taxon.TaxonGenerator'>, <class 'h1_hesc.kb_gen.compartments.CompartmentsGenerator'>, <class 'h1_hesc.kb_gen.chromosomes.ChromosomesGenerator'>, <class 'h1_hesc.kb_gen.genes_transcripts_exons_proteins.GenesTranscriptsExonsProteinsGenerator'>, <class 'h1_hesc.kb_gen.regulatory_modules.RegulatoryModulesGenerator'>, <class 'h1_hesc.kb_gen.metabolites.MetabolitesGenerator'>, <class 'h1_hesc.kb_gen.properties.PropertiesGenerator'>, <class 'h1_hesc.kb_gen.transcripts.TranscriptsGenerator'>, <class 'h1_hesc.kb_gen.metabolic_network.MetabolicNetworkGenerator'>, <class 'h1_hesc.kb_gen.proteins.ProteinsGenerator'>, <class 'h1_hesc.kb_gen.complexes.ComplexesGenerator'>, <class 'h1_hesc.kb_gen.metabolic_reaction_kinetics.MetabolicReactionKineticsGenerator'>, <class 'h1_hesc.kb_gen.protein_modifications.ProteinModificationsGenerator'>)[source]¶

clean_and_validate_options()[source]¶: Apply default options and validate options

run()[source]¶: Generate a knowledge base of experimental data for a whole-cell model :returns: knowledge base :rtype: wc_kb.core.KnowledgeBase

h1_hesc.kb_gen.core.main(core_path=None, seq_path=None, read=False, test_kb=None, test_mode=False, in_phase=False)[source]¶

Generates a knowledge base for H1 hESCs

Parameters

core_path (str, optional) – path to read/write the knowledge base
seq_path (str, optional) – path to the genome sequence
read (bool, optional) – if True, the knowledge base will be generated by deserializing a previously created excel spreadsheet given by the core_path, else, it will be generated by running the generator and serializing the output
test_kb (wc_kb.core.KnowledgeBase, optional) – KB object provided for test purpose only
test_mode (bool, optional) – if True, the knowledge base generator will run in test mode using a subset of the genome
in_phase (bool, optional) – if True, the knowledge base generator will run in phase and write to the path between phases

3.1.1.2.6. h1_hesc.kb_gen.genes_transcripts_exons_proteins module¶

Reconstruct genes, pre-rnas, transcripts, exons, CDSs and proteins

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Author: Jonathan Karr <jonrkarr@gmail.com>
Date: 2018-01-30
Copyright: 2018, Karr Lab
License: MIT

class h1_hesc.kb_gen.genes_transcripts_exons_proteins.GenesTranscriptsExonsProteinsGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Create genes, pre-rnas, transcripts, exons, CDSs and proteins for the knowledge base by parsing genome annotation file. To speed processing, the genome annotation file is divided into smaller chunks that is further divided into batches. The batches are processed sequentially while the chunks in each batch are processed in parallel. At the end of each batch, the intermediate results will be dumped into a pickle file.

clean_and_validate_options()[source]¶

Apply default options and validate options

Options:

data_path (str, optional): path to genome annotation lifted from reference genome,
default is system path where an annotation file will be downloaded from Quilt
trna_data_path (str, optional): path to data on cytoplasmic tRNA
mt_trna_data_path (str, optional): path to data on mitochondrial tRNA
rrna_list (list of str, optional): list of gene name of each
representative nuclear-encoded rRNA in the genome annotation
rrna_18S_28S_list (list of str, optional): list of transcript name of
representative 18S and 28S rRNAs not in the genome annotation
pickle_path (str): path to directory to store pickle files of constructed genes,
pre-rnas, transcripts, exons, CDSs and proteins
id_conversion_path (str): path of gene-transcript-protein id conversion table
chunk_size (int, optional): byte size of each chunk when breaking genome annotation
file into smaller chunks, default is 1024*1024
batch_size (int, optional): number of chunks in each batch, default is 300
test_mode (bool, optional): mode for testing and debugging where
only the first batch will be processed if set to True, default is False
scaled_down_gene_list (:obk:`list`, optional): list of ensembl protein-producing gene IDs
that if provided, will be the only genes constructed
scaled_down_trna_list (:obk:`list`, optional): list of nuclear-encoded tRNA gene IDs
that if provided, will be the only nuclear-encoded tRNA genes constructed

gen_components()[source]¶: Construct genes, pre-rnas, exons, transcripts and proteins for the knowledge base

get_data()[source]¶: Get tRNA annotation for the knowledge base

process_data()[source]¶: Read and process the data

3.1.1.2.7. h1_hesc.kb_gen.metabolic_network module¶

Reconstruct the metabolic network

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2018-08-10
Copyright: 2018, Karr Lab
License: MIT

class h1_hesc.kb_gen.metabolic_network.MetabolicNetworkGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Reconstruct metabolic network for the knowledge base

Starting from Recon 2.2, existing reactions are modified and new reactions and their associated genes are added to the network to:

transport nutrients from the media, mTeSR1, that is commonly use in hESC cultures

produce and recycle metabolites that are involved in non-metabolic pathways

serve as the alternative pathways to non-essential reactions

improve the representation of energy metabolism by modifying mitochondrial metabolism in Recon 2.2 based on MitoCore

Options:

metabolic_network_path (str): path of human metabolic reconstruction
new_reactions_data_path (str): path of new reactions and modified reactions
scaled_down_reaction_list (:obk:`list`, optional): list of metabolic reaction IDs
that if provided, will be the only metabolic reactions constructed

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct metabolic reactions for the knowledge base

get_data()[source]¶: Get human metabolic network and manually curated reactions from input files

3.1.1.2.8. h1_hesc.kb_gen.metabolic_reaction_kinetics module¶

Retrieve kinetic data from SABIO-RK for a taxon such as mammals (40674)

Author: Jonathan Karr <jonrkarr@gmail.com>
Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2017-07-25
Copyright: 2018, Karr Lab
License: MIT

class h1_hesc.kb_gen.metabolic_reaction_kinetics.MetabolicReactionKineticsGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Reconstruct the rate laws and rate parameters of metabolic reactions from Sabio-RK

First, the InChI structures of metabolites are used to find the sabio compound for each species. Then, the EC numbers of reactions that are missing are imputed. Data measured in organisms with the same taxon as the modeled cell are retrieved from Sabio-RK and matched to the reactions. Each kinetic parameter associated to a reaction is then matched to each protein complex.

Options:

id_conversion_path (str): path of gene-transcript-protein id conversion table
mutant_inclusion (bool): mutant inclusion parameter for querying Sabio-RK
min_temp (float): minimum temperature for querying Sabio-RK
max_temp (float): maximum temperature for querying Sabio-RK
min_ph (float): minimum pH for querying Sabio-RK
max_ph (float): maximum pH for querying Sabio-RK
j_threshold (float): the minimum Jaccard similarity between the gene-reaction-rules in metabolic network
and the reconstructed complexes that determines if a protein complex is assign to the reaction

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct metabolic reaction kinetic laws and parameters for the knowledge base

get_data()[source]¶: Get kinetic laws and rate parameters from Sabio-RK

process_data()[source]¶

Match reaction kinetics to complexes

First, a Jaccard similarity is used to compare and assign reconstructed protein complexes to each reaction. Then, a protein complex is assigned kinetic data whose enzyme subunits are subset of the protein complex.

3.1.1.2.9. h1_hesc.kb_gen.metabolites module¶

Reconstruct the metabolites in a cell

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2018-08-05
Copyright: 2018, Karr Lab
License: MIT

class h1_hesc.kb_gen.metabolites.MetabolitesGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Creates metabolites for the knowledge base from input files.

Options: * metabolic_network_path (str): path of human metabolic reconstruction * new_metabolites_data_path (str): path of manually curated new metabolites * concentration_data_path (str): path of manually curated metabolite concentrations * scaled_down_metabolite_list (:obk:`list`, optional): list of metabolite IDs

that if provided, will be the only metabolites whose free-pool concentrations are constructed

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct metabolite species types and concentrations for the knowledge base

get_data()[source]¶

Get metabolite species from human metabolic network and manually added metabolite species from input file
Get intracellular concentrations of metabolites from input file and datanator

process_data()[source]¶

Merge and assign the concentration data to the metabolite species type in the knowledge base using the following steps: 1) Assign concentration data from manual curation to the metabolite species 2) For the remaining metabolite species:

Exactly match the InChI strings of metabolites in the database

If no match is found in step a, match InChI string by successively ignoring
charge and stereochemistry layers

If no match is found in step b, extract non-matching InChI string that are not
cardiolipin for manual matching

3.1.1.2.10. h1_hesc.kb_gen.properties module¶

Reconstruction of cell properties

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2018-11-15
Copyright: 2018, Karr Lab
License: MIT

class h1_hesc.kb_gen.properties.PropertiesGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Create cell properties for the knowledge base

Options:

data_path (dict): path to data of cell properties

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct cell properties for the knowledge base

get_data()[source]¶: Get data for cell properties

process_data()[source]¶: Process data for knowledge base components

3.1.1.2.11. h1_hesc.kb_gen.protein_modifications module¶

Phosphorylation sites and ratio calculation: from the H1-ESC iTRAQ-8plex data (Phanstiel et al Nat Methods 2011)

Author: Yassmine Chebaro <chebaro@igbmc.fr>
Date: 2019-01-24
Copyright: 2019, Karr Lab
Licence: MIT

class h1_hesc.kb_gen.protein_modifications.ProteinModificationsGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Create phosphorylation states for the knowledge base

The ratio of phosphorylation for each protein is calculated using the quantification provided from Phanstiel et al supp mat, and is the following:

quantification(phos)/(quantification(proteome)+quantification(phos))

In cases where quantification of a protein is NA or absent, no ratio for the corresponding phospho-protein is calculated.

Options:

phosphoproteomics_data_path (str): path of the phosphoproteome data from Phanstiel et al
id_conversion_path (str): path of gene-transcript-protein id conversion table

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct protein modifications for the knowledge base

get_data()[source]¶: Get proteome and phosphoproteome data from xlsx spreadsheets and get IPI mapping to ensembl protein

process_data()[source]¶

Phosphorylation ratio per site for each protein are calculated from the MS quantification data obtained from Phanstiel et al Nat Methods 2011. The following steps are performed to allow for more flexibility in the future if needed (what to do with N/A or absent data)

1) Extract phosphorylated proteins when non-phosphorylated equivalent is present in proteomics data and calculate the ratio for each phospho-site per protein per set of experiment. Three replicates are available for H1 data: ES (H1) iTRAQ-117, ES (H1) iTRAQ-113 and ES (H1) iTRAQ-117 (changed to ES (H1) iTRAQ-117b here for simplification). Clean the ratios: where N/A or no value is present the previous step will output nan, these ratios are replaced with nan. This is not combined with the previous step in case one does not want to remove the nans

2) Calculate average ratio per site per protein and map the IPI to ensembl protein after removing the nan ratios

3.1.1.2.12. h1_hesc.kb_gen.proteins module¶

Reconstruction of proteins

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2017-07-19
License: MIT

class h1_hesc.kb_gen.proteins.ProteinsGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Create proteins for the knowledge base

First, protein species in each compartment is generated by merging data of protein localization from metabolic network and Cell Atlas into a consensus set. For transporter proteins in the metabolic network that catalyze reactions involving metabolites from multiple compartments, they are localized to the membrane compartments to ensure the reactions will occur in the model. For proteins not included in the metabolic network and Cell Atlas, localization data from Uniprot are used. For proteins not included in all data sources above, they are assumed to localize to the cytosol.

Second, H1 protein abundance data are extracted from PaxDB and supplemented by merging with protein abundance data from other human cell lines. In cases where multiple supplementary datasets contain measurement for the same protein, the supplementary dataset that correlates best with the main dataset will be selected. A housekeeping protein is used to normalize the datasets. The abundance of proteins without any measurement are set to the median measured abundances. Inconsistencies between the proteomic and transcriptomic data are then resolved.

Third, the concentration of each protein species in each compartment is set by distributing whole-cell protein concentration based on the volumetric fraction of the compartment.

Fourth, the half-lives of proteins are taken from the literature. In cases where measurements are not available, the half-lives are estimated based on the correlation between half-life and protein abundance that have been reported in the literature.

Options:

id_conversion_path (str): path to gene-transcript-protein id conversion table
localization_data_path (str): path to localization data from Cell Atlas
uniprot_localization_data_path (str): path to localization data from Uniprot
substructure_to_compartment (dict): dictionary to translate subcellular
structures defined in Cell Atlas (keys) to compartments defined in Recon 2.2
uniprot_location_to_compartment (dict): dictionary translate subcellular
locations defined in Uniprot (keys) to compartments defined in Recon 2.2
housekeeping_gene (str): the Ensembl Protein id of the housekeeping protein
main_dataset (str): the filename of H1 dataset in PaxDB
half_life_data_path (str): path to protein half-life data
rrna_18S_seq_path (str, optional): path to sequence fasta file for representative
18S rRNA
rrna_28S_seq_path (str, optional): path to sequence fasta file for representative
28S rRNA

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct protein concentrations and half-lives for the knowledge base

get_data()[source]¶: Get protein localization data from Cell Atlas, protein abundance data from PaxDB and protein half-life data

process_data()[source]¶

Protein localisation data from Cell Atlas, metabolic network and Uniprot are merged

Whole-cell absolute protein abundances (mol/cell) are calculated using the following steps: 1) Combine relative abundances measured in H1 with data lifted over from

other cell line

Set the relative abundances of non-measured proteins to the median protein
abundance reconstructed in Step 1.
Resolve inconsistencies between the reconstructed relative protein abundances
and observed transcript abundances: a) If the relative protein abundance is zero but the transcript has been shown to

be expressed in H1, set the protein abundance to be the median abundance reconstructed in Step 1
1. If the relative protein abundance is not zero but the measured transcript
  abundance is zero, check whether the proteomic data was measured in H1 or lifted from another cell line. If the protein abundance was lifted from another cell line, set the protein abundance to zero. Otherwise, set the transcript abundance to the median transcript abundance
Calculate absolute protein abundances by multiplying relative protein abundances to
the total H1 protein mass

Because each protein defined above represents one generic product of each gene, its abundance is further distributed among its protein variants based on the relative abundance of the transcript that is translated into each variant.

The correlation between protein half-life and protein abundance are then determined and used to estimate the half-life of proteins where measurements are not available.

3.1.1.2.13. h1_hesc.kb_gen.regulatory_modules module¶

Reconstruct regulatory features

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2018-11-27
License: MIT

class h1_hesc.kb_gen.regulatory_modules.RegulatoryModulesGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Create regulatory elements and modules for the knowledge base

clean_and_validate_options()[source]¶

Apply default options and validate options

Options:

reg_activity_data_path (str): path to data on regulatory features and activities
motif_feature_data_path (str): path to data on motif features

gen_components()[source]¶: Construct regulatory elements for the knowledge base

get_data()[source]¶: Get data for the knowledge base

process_data()[source]¶: Read and process the data

3.1.1.2.14. h1_hesc.kb_gen.taxon module¶

Reconstruct the taxon

Author: Jonathan Karr <jonrkarr@gmail.com>
Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2018-02-05
License: MIT

class h1_hesc.kb_gen.taxon.TaxonGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Generator for the taxon of the organism

Options:

scientific_name (str): the scientific name of organism

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct taxon for the knowledge base

get_data()[source]¶: Get taxon id of organism

3.1.1.2.15. h1_hesc.kb_gen.transcripts module¶

Reconstruction of transcripts

Author: Yin Hoon Chew <yinhoon.chew@mssm.edu>
Date: 2017-08-29
License: MIT

class h1_hesc.kb_gen.transcripts.TranscriptsGenerator(knowledge_base, options=None)[source]¶

Bases: wc_kb_gen.core.KbComponentGenerator

Create transcripts for the knowledge base

The copy number of each transcript species type is calculated by dividing RNA biomass by the sum of the products of the relative abundance and molecular weight of each RNA, and multiplying by the relative abundance of the RNA in TPM (transcript per million).

The half-life (in seconds) of each transcript species type is reconstructed based on its Gene Ontology (biological process) by referring to the correlation published in Yang et al (2003). The average value is used if a transcript species type is associated with more than one GO. If no association is found, a median half-life of 10 hrs from BioNumbers is used.

Options:

data_path (str): path to transcriptomic data
trna_data_path (str, optional): path to tRNA
rrna_variants_path (str, optional): path to directory where pickle file
of extra rrna variants is stored
rrna_18S_seq_path (str, optional): path to sequence fasta file for representative
18S rRNA
rrna_28S_seq_path (str, optional): path to sequence fasta file for representative
28S rRNA
correlation_path (str): path to correlation between half-life and GO
id_conversion_path (str): path of gene-transcript-protein id conversion table

clean_and_validate_options()[source]¶: Apply default options and validate options

gen_components()[source]¶: Construct transcript concentrations for the knowledge base

get_data()[source]¶: Get transcript abundance data in TPM and the correlation between GO and half-life from input files

process_data()[source]¶: Calculate the copy number and estimate the half-life of each transcript

3.1.1.2.16. Module contents¶

Reconstruction building API

Author: Jonathan Karr <jonrkarr@gmail.com>
Date: 2018-01-30
License: MIT