Edinburgh-Genome-Foundry/BandWitch

💫 Computer-aided DNA assembly validation and identification from restriction digests.

BandWitch

BandWitch (full documentation here) is a Python library for the planning and analysis of restriction experiments in DNA assembly operations. BandWitch implements methods for selecting the best enzyme(s) to validate or identify DNA assemblies. It also provides report generation methods to automatically validate/identify assemblies from experimental data.

You can try BandWitch's enzyme suggestion feature in this web demo, and the sequence validation (from AATI fragment analyzer files) in this other demo.

Installation

You can install BandWitch through PIP:

pip install bandwitch

On Ubuntu, you may need to install libblas first:

sudo apt-get install libblas-dev liblapack-dev

Alternatively, you can unzip the source files in a folder and type:

python setup.py install

Enzyme selection with BandWitch

In the following examples, we assume that we have a set of 12 constructs which we will need to either validate (i.e. we digest these constructs and compare each pattern with the expected pattern for that construct) or identify (i.e. we will digest an a priori unknown construct and use the migration patterns to unambiguously identify each construct among the 12 possible candidates).

For validation purposes, the difficulty is to find a digestion that will produce harmonious patterns for all the constructs at once: well-spaced bands, and not too many or too few of them. For identification purposes, the difficulty is to find a digestion giving very distant patterns for each construct in the set of candidates.

Every time when the problem cannot be solved with a single digestion, BandWitch can propose 2 or 3 digestions which collectively solve the problem.

Important: when providing BandWitch with a Biopython record, make sure to set the topology, defined by: record.annotations['topology'] = 'linear'|'circular'

Finding enzymes that "work well" for many constructs

Here is the code to select enzymes that will produce nice patterns for all constructs, for validation:

from bandwitch import IdealDigestionsProblem, LADDERS, load_record

# DEFINE THE SEQUENCES AND THE ENZYME SET
enzymes = ["EcoRI", "BamHI", "XhoI", "EcoRV", "SpeI", "XbaI",
           "NotI", "SacI", "SmaI", "HindIII", "PstI"]
sequences = [
    load_record(record_file=f, topology='circular')
    for f in some_list_of_genbank_files
]

# SELECT THE BEST SINGLE DIGESTION WITH AT MOST ENZYMES
problem = IdealDigestionsProblem(enzymes=enzymes,
                                 ladder=LADDERS['100_to_4k'],
                                 sequences=sequences,
                                 max_enzymes_per_digestion=2)
score, selected_digestions = problem.select_digestions(max_digestions=1)

# PLOT THE BAND PATTERNS PRODUCED BY THE SELECTED DIGESTION
problem.plot_digestions(
    digestions=selected_digestions,
    patterns_props={'label_fontdict': {'rotation': 35}},
    target_file="ideal_digestions.png"
)

Result:

Finding enzymes that will differentiate many constructs

To select enzymes that will produce different patterns for each construct, for identification:

from bandwitch import (SeparatingDigestionsProblem, list_common_enzymes,
                       LADDERS, load_record)


# DEFINE SEQUENCES AND ENZYME SET (6-CUTTERS WITH >3 COMMERCIAL PROVIDERS)
enzymes = list_common_enzymes(site_length=(6,), min_suppliers=3)
sequences = [
    load_record(genbank_file_path, topology='circular')
    for genbank_file_path in some_list_of_genbank_files
]

# SELECT THE BEST DIGESTION PAIRS (AT MOST 1 ENZYME PER DIGESTION)
problem = SeparatingDigestionsProblem(enzymes=enzymes,
                                      ladder=LADDERS['100_to_4k'],
                                      sequences=sequences,
                                      max_enzymes_per_digestion=1)
score, selected_digestions = problem.select_digestions(max_digestions=2)

# GENERATE A FIGURE OF THE BAND PATTERNS
problem.plot_digestions(
    selected_digestions,
    patterns_props={'label_fontdict': {'rotation': 35}},
    target_file="separating_digestions.png"
)

problem.plot_distances_map(digestions=selected_digestions,
                           target_file="separating_digestions_distances.png")

Result:

In the result above, each construct has a unique "fingerprint". Assuming that you have an unlabelled DNA sample which could be any of these assemblies, then simply digesting the sample with MspA1I and BsmI will give you 2 patterns which collectively will correspond to a unique assembly.

Usage: Construct validation or identification from experimental data

This part is still under construction.

BandWitch can process output files from an automated fragment analyzer and produce informative reports as illustrated below:

License = MIT

BandWitch is an open-source software originally written at the Edinburgh Genome Foundry by Zulko and released on Github under the MIT license (Copyright 2017 Edinburgh Genome Foundry). Everyone is welcome to contribute!

More biology software

BandWitch is part of the EGF Codons synthetic biology software suite for DNA design, manufacturing and validation.