Usage

import combinatorial_peptide_pooling as cpp

To use the package for basic tasks, the Quickstart section is enough. To read more about used functions, check other sections.

Quickstart

import combinatorial_peptide_pooling as cpp

# number of pools
n_pools = 12
# peptide occurrence
iters = 4
# number of peptides
len_lst = 253

# address arrangemement
b, lines = cpp.address_rearrangement_AU(n_pools=n_pools, iters=iters, len_lst=len_lst)

# add your peptides to lst
lst = list(pd.read_csv('peptides.csv', sep = "\t"))

# pooling scheme generation
pools, peptide_address = cpp.pooling(lst=lst, addresses=lines, n_pools=n_pools)

# simulation
check_results = cpp.run_experiment(lst=lst, peptide_address=peptide_address, ep_length=8, pools=pools, iters=iters, n_pools=n_pools, regime='without dropouts')

# STL files generation
# add peptide scheme to peptides_table_stl, with header and index as column and row numbers
peptides_table_stl = pd.read_csv('peptides_scheme.tsv', sep = "\t", index_col = 0)
pools_df = pd.DataFrame({'Peptides': [';'.join(val) for val in pools.values()]}, index=pools.keys())
meshes_list = cpp.pools_stl(peptides_table = peptides_table_stl, pools = pools_df, rows = 16, cols = 24, length = 122.10, width = 79.97,
           thickness = 1.5, hole_radius = 2, x_offset = 9.05, y_offset = 6.20, well_spacing = 4.5)
cpp.zip_meshes_export(meshes_list)

More detailed quickstart

  1. (Optional) Check your peptide list for overlap consistency.

    Note

    Incosistent overlap length can lead to hindered results interpretation.

    You can check all peptides for their overlap length with the next peptide (list of peptides should be ordered):

    cpp.all_overlaps(lst) Counter object
    Parameters:

    lst (list) – ordered list of peptides

    Returns:

    Counter object with the dictionary, where the key is the overlap length and the value is the number of pairs with such overlap.

    Return type:

    Counter object

    >>> cpp.all_overlaps(lst)
    Counter({12: 251, 16: 1})
    

    => 251 pairs of peptides with an overlap of length of 12 amino acids, and 1 pair with an overlap of length 16 amino acids.

    Also, you can check which peptides have such an overlap with the next peptide:

    cpp.find_pair_with_overlap(lst, target_overlap) list
    Parameters:
    • lst (list) – ordered list of peptides

    • target_overlap (int) – overlap length

    Returns:

    list of lists with peptides with specified overlap length.

    Return type:

    list

    >>> cpp.find_pair_with_overlap(lst, 16)
    [['FDEDDSEPVLKGVKLHY', 'DEDDSEPVLKGVKLHYT']]
    

    => Overlap of length 16 amino acids is in peptides FDEDDSEPVLKGVKLHY and DEDDSEPVLKGVKLHYT.

    Also, you can check what number of peptides share the same epitope. It might help to interpret the results later.

    cpp.how_many_peptides(lst, ep_length) Counter object, dictionary
    Parameters:
    • lst (list) – ordered list of peptides

    • ep_length (int) – expected epitope length

    Returns:

    1. the Counter object with the number of epitopes shared across the number of peptides;

    2. the dictionary with all possible epitopes of expected length as keys and the number of peptides where these epitopes are present as values.

    Return type:

    Counter object, dictionary

    >>> t, r = cpp.how_many_peptides(lst, 8)
    >>> t
    Counter({1: 6, 2: 1256, 3: 4})
    >>> r
    {'MFVFLVLL': 1,'FVFLVLLP': 1,VFLVLLPL': 1,'FLVLLPLV': 1,'LVLLPLVS': 1,'VLLPLVSS': 2, ...,}
    

    => There are 6 epitopes present in a single peptide, 1256 epitopes present shared by two peptides, and 4 epitopes shared by 4 peptides. For each epitope, number of peptides sharing it is in the dictionary.

  2. (Optional) Then you need to determine peptide occurrence across pools, i.e. to how many pools one peptide would be added.

    Note

    Peptide occurrence affects number of peptides in one pool, and therefore too high peptide occurrence may lead to higher dilution of a single peptide.

    cpp.find_possible_k_values(n, l) list
    Parameters:
    • n (int) – number of pools

    • l (int) – number of peptides

    Returns:

    list with possible peptide occurrences given number of pools and number of peptides.

    Return type:

    Counter object, dictionary

    >>> cpp.find_possible_k_values(12, 250)
    [4, 5, 6, 7, 8]
    

    => Given 12 pools and 250 peptides, you can use peptide occurrence equal to 4, 5, 6, 7, 8.

    Choose one occurrence value appropriate for your task and proceed.

  3. Now, you need to find the address arrangement given your number of pools, number of peptides, and peptide occurrence.

    We suggest you use the cpp.address_rearrangement_AU() function. In the section Address arrangement you can find other functions that can perform such a task (based on Gray codes and on a trivial Hamiltonian path search).

    Note

    With large parameters, the algorithm needs some time to finish the arrangement. If the arrangement fails, try with other parameters.

    cpp.address_rearrangement_AU(n_pools, iters, len_lst) list, list
    Parameters:
    • n_pools (int) – number of pools

    • iters (int) – peptide occurrence

    • len_lst (int) – number of peptides

    Returns:

    1. list with number of peptides in each pool;

    2. list with address arrangement

    Return type:

    list, list

    >>> cpp.address_rearrangement_AU(n_pools=12, iters=4, len_lst=250)
    >>> b
    [81, 85, 85, 85, 81, 82, 87, 81, 85, 81, 84, 83]
    >>> lines
    [[0, 1, 2, 3],[0, 1, 3, 6],[0, 1, 6, 8],[1, 6, 8, 9],[6, 8, 9, 11], ... ]
    

    => You will get the expected number of peptides in each pool and address arrangement, which will be used in following steps.

  4. Now, you can distribute peptides across pools using the produced address arrangement. One peptide will be added to one produced address.

    Note

    Keep in mind that peptides should be ordered as they overlap.

    cpp.pooling(lst, addresses, n_pools) dictionary, dictionary
    Parameters:
    • lst (list) – ordered list with peptides

    • addresses (list) – produced address arrangement

    • n_pools (int) – number of pools

    Returns:

    1. pools – dictionary with keys as pools indices and values as peptides that should be added to this pools;

    2. peptide address – dictionary with peptides as keys and corresponding addresses as values.

    Return type:

    dictionary, dictionary

    >>> pools, peptide_address = cpp.pooling(lst=lst, addresses=lines, n_pools=12)
    >>> pools
    {0: ['MFVFLVLLPLVSSQCVN','VLLPLVSSQCVNLTTRT',VSSQCVNLTTRTQLPPA', ...], 1: ['MFVFLVLLPLVSSQCVN','VLLPLVSSQCVNLTTRT','TQDLFLPFFSNVTWFHA', ...], ... }
    >>> peptide_address
    {'MFVFLVLLPLVSSQCVN': [0, 1, 2, 3], 'VLLPLVSSQCVNLTTRT': [0, 1, 2, 10], ... }
    

    => You will get the pooling scheme and peptide addresses.

  5. Now, you can run the simulation using produced pools and peptide_address.

    The simulation produces a DataFrame with every possible epitope of the provided length and all pools where this epitope is present. This table is needed to interpret the results.

    The function has two regimes: with and without drop-outs. Without drop-outs, it returns a table as there were no mistakes, and all pools that should be activated were activated. With drop-outs, it returns a table with all possible mistakes (i.e.all possible non-activated pools). This option will need time to be generated, usually several minutes, although it depends on the number of peptides and on occurrence.

    cpp.run_experiment(lst, peptide_address, ep_length, pools, iters, n_pools, regime) pandas DataFrame

    Note

    Simulation may take several minutes, especially upon “with drop-outs” regime.

    Parameters:
    • lst (list) – ordered list with peptides

    • peptide_address (dictionary) – peptides addresses produced by pooling

    • ep_length (int) – expected epitope length

    • pools (dictionary) – pools produced by pooling

    • iters (int) – peptide occurrence

    • n_pools (int) – number of pools

    • regime (“with dropouts” or “without dropouts”) – regime of simulation, with or without drop-outs

    Returns:

    1. pools – dictionary with keys as pools indices and values as peptides that should be added to this pools;

    2. peptide address – dictionary with peptides as keys and corresponding addresses as values.

    Return type:

    dictionary, dictionary

    >>> df = cpp.run_experiment(lst=lst, peptide_address=peptide_address, ep_length=8, pools=pools, iters=iters, n_pools=n_pools, regime='without dropouts')
    
    >>> df
    

    Peptide

    Address

    Epitope

    Act Pools

    # of pools

    # of epitopes

    # of peptides

    Remained

    # of lost

    Right peptide

    Right epitope

    MFVFLVLLPLVSSQCVN

    [0, 1, 2, 3]

    MFVFLVLL

    [0, 1, 2, 3]

    4

    5

    1

    0

    True

    True

    MFVFLVLLPLVSSQCVN

    [0, 1, 2, 3]

    MFVFLVLL

    [0, 1, 2, 3]

    4

    5

    1

    0

    True

    True

    MFVFLVLLPLVSSQCVN

    [0, 1, 2, 3]

    VLLPLVSS

    [0, 1, 2, 3, 10]

    5

    5

    2

    0

    True

    True

    VLLPLVSSQCVNLTTRT

    [0, 1, 2, 10]

    VLLPLVSS

    [0, 1, 2, 3, 10]

    5

    5

    2

    0

    True

    True

    Peptide — peptide sequence

    Address — pool indices where this peptide should be added

    Epitope — checked epitope from this peptide

    Act pools — list with pool indices where this epitope is present

    # of pools — number of pools where this epitope is present

    # of epitopes — number of epitopes that are present in the same pools (= number of possible peptides upon activation of such pools)

    # of peptides — number of peptides in which there are epitopes that are present in the same pools (= number of possible peptides upon activation of such pools)

    Remained — only upon regime=”with dropouts”, list of pools remained after mistake

    # of lost — only upon regime=”with dropouts”, number of dropped pools due to mistake

    Right peptide — True or False, whether the peptide is present in the list of possible peptides

    Right epitope — True or False, whether the peptide is present in the list of possible peptides

    To interpret the results of the experiment, you need to find all rows where the “Act Pools” column contains your combination of activated pools. Then, you will know all possible peptides and epitopes that could lead to the activation of such a combination of pools.

    If you can not find your combination of activated pools in the table, here is the sequence of actions.

    After the experiment, you will know the number of activated pools. This number depends on the length of overlap and the length of the expected epitope. You can check the distribution of epitope presence in your peptides using cpp.how_many_peptides() function. The number of activated pools would be equal to peptide occurrence plus one per additional peptide sharing this epitope.

    This way, if the epitope is present only in 1 peptide (usually, it is the case for epitopes at the ends of the protein), then the number of activated pools is equal to peptide occurrence. If the epitope is present in two peptides, then the number of activated pools is equal to peptide occurrence +1.

    If overlap length is consistent across all peptides, then the number of activated pools would be the same for almost all epitopes (except for the epitopes at the ends of the protein). Although even if the overlap is inconsistent, you can use the analysis, but it will hinder the interpretation of the results in some cases.

    If a shift length between two peptides is equal to or less than the expected epitope length divided by two, then the number of activated pools should be equal to the peptide occurrence value + 1.

    If the number of activated pools is less than according to the rule described above, then three options are possible:

    • The target peptide is the peptide at the end of your peptide list, and the target epitope is located not in an overlap of this peptide with the next one. This could be checked easily: if your activated pools are not the same as the activated pools for any epitope from the first or last peptide, then you should check our second option.

    • For the target peptide, overlap with its neighbor is less than usual, and therefore target epitope is not shared by the usual number of peptides. You can check that using cpp.all_overlaps() or cpp.how_many_peptides(). Nevertheless, given the absence of drop-outs, you still should be able to find the target peptide in the table with simulation results by searching for all rows where the “Act Pools” column contains your combination of activated pools.

    • Some pools were not activated, although they should be; then, we recommend using the “with drop-outs” regime of the simulation. It imitates drop-outs of all possible pools, so you should be able to find your case in the resulting table.

    If the number of activated pools is higher than according to the rule described above, then two options are possible:

    • For the target peptide, overlap with its neighbor is bigger than usual, and therefore target epitope is shared between more peptides. You can check that using cpp.all_overlaps() or cpp.how_many_peptides(). Nevertheless, given the absence of drop-outs, you still should be able to find the target peptide in the table with simulation results by searching for all rows where the “Act Pools” column contains your combination of activated pools.

    • Some pools were activated, although they should not be. This issue is not addressed in the package.

    >>> df = cpp.run_experiment(lst=lst, peptide_address=peptide_address, ep_length=8, pools=pools, iters=iters, n_pools=n_pools, regime='with dropouts')
    >>> df
    

    Peptide

    Address

    Epitope

    Act Pools

    # of pools

    # of epitopes

    # of peptides

    Remained

    # of lost

    Right peptide

    Right epitope

    MFVFLVLLPLVSSQCVN

    [0, 1, 2, 3]

    MFVFLVLL

    [0, 1, 2, 3]

    4

    40

    12

    [0, 1, 2]

    1

    True

    False

    MFVFLVLLPLVSSQCVN

    [0, 1, 2, 3]

    MFVFLVLL

    [0, 1, 2, 3]

    4

    76

    25

    [0, 1, 3]

    1

    True

    False

    RTQLPPAYTNSFTRGVY

    [8, 9, 10, 11]

    RTQLPPAY

    [0, 8, 9, 10, 11]

    5

    5

    2

    [0, 8, 9, 10, 11]

    0

    True

    True

    RTQLPPAYTNSFTRGVY

    [8, 9, 10, 11]

    TQLPPAYT

    [0, 8, 9, 10, 11]

    5

    190

    53

    [8, 9]

    3

    True

    True

    Peptide — peptide sequence

    Address — pool indices where this peptide should be added

    Epitope — checked epitope from this peptide

    Act pools — list with pool indices where this epitope is present

    # of pools — number of pools where this epitope is present

    # of epitopes — number of epitopes that are present in the same pools (= number of possible peptides upon activation of such pools)

    # of peptides — number of peptides in which there are epitopes that are present in the same pools (= number of possible peptides upon activation of such pools)

    Remained — only upon regime=”with dropouts”, list of pools remained after mistake

    # of lost — only upon regime=”with dropouts”, number of dropped pools due to mistake

    Right peptide — True or False, whether the peptide is present in the list of possible peptides

    Right epitope — True or False, whether the peptide is present in the list of possible peptides

    Right peptide and Right epitope columns are needed to check the algorithm of dropped pool recovery. Either “Right peptide” or “Right epitope” should contain the value “True”; otherwise, recovery was unsuccessful.

    Also, the regime “with drop-outs” can not differentiate between dropped pools due to a mistake and absent pools due to experiment design. This way, for epitopes located at the end of proteins, the algorithm would think that pools were dropped and would try to recover them. Because of that, if you suspect the epitope located at the end of the peptide to be the target epitope, we recommend first using the “without drop-outs” regime. You can look at the sequence of actions described above. The same applies to peptides with longer overlap. So, we strongly recommend using peptides with consistent overlap length.

  6. (Optional) To avoid mixing pools manually, you can print special punch cards using files with their 3D models produced by this step.

    One punch card is needed for each pool. Each punch card is a thin card with holes located at the spots where the needed peptides are located in the plate. Therefore, each punch card has the number of holes equal to the number of peptides in a pool. Then, this card should be placed on an empty tip box, and a tip should be inserted into each hole. This way, if you are using a multichannel pipette, all tips are already arranged to take only the required peptides.

    [The process you can look up here.]

    To generate the files with 3D models, you need two functions.

    Note

    The rendering of 3D models is a long process, so it could take time.

    cpp.pools_stl(peptides_table, pools, rows=16, cols=24, length=122.10, width=79.97, thickness=1.5, hole_radius=4.0 / 2, x_offset=9.05, y_offset=6.20, well_spacing=4.5) dictionary
    Parameters:
    • peptides_table (pandas DataFrame) – table representing the arrangement of peptides in a plate, is not produced by any function in the package

    • pools (pandas DataFrame) – table with a pooling scheme, where one row represents each pool, pool index is the index column, and a string with all peptides added to this pool separated by “;” is “Peptides” column.

    • rows (int) – int

    • cols (int) – number of columns in your plate with peptides

    • length (float) – length of the plate in mm

    • width (float) – width of the plate in mm

    • thickness (float) – desired thickness of the punch card, in mm

    • hole_radius (float) – the radius of the holes, in mm, should be adjusted to fit your tip

    • x_offset (float) – the margin along the X axis for the A1 hole, in mm

    • y_offset (float) – the margin along the Y axis for the A1 hole, in mm

    • well_spacing (float) – the distance between wells, in mm

    Returns:

    dictionary with Mesh objects, where key is pool index, and value is a Mesh object of a corresponding punch card.

    Return type:

    dictionary

    >>> meshes_list = cpp.pools_stl(peptides_table, pools, rows = 16, cols = 24, length = 122.10, width = 79.97, thickness = 1.5, hole_radius = 2.0, x_offset = 9.05, y_offset = 6.20, well_spacing = 4.5)
    

    Now, you need to pass generated dictionary to the function exporting it as a .zip file.

    cpp.zip_meshes_export(meshes_list) None
    Parameters:

    meshes_list (dictionary) – dictionary with Mesh objects, generated in previous step

    Returns:

    export Mesh objects as STL files in .zip archive.

    Return type:

    None

    >>> cpp.zip_meshes_export(meshes_list)
    

    => You will get a .zip archive with generated STL files. Then, you can send these STL files directly to a 3D printer. We recommend writing the index of the pool on the punch card. Also, you can check the generated STL files using OpenSCAD.

Address arrangement

Note

Method for n-bit balanced Gray code construction is based on the textbook Counting sequences, Gray codes and lexicodes. Method for construction of balanced Gray code with flexible length is based on the paper Balanced Gray Codes With Flexible Lengths.

cpp.find_q_r(n) tuple
Parameters:

n (int) – number

Returns:

solution for the equation 2**n = n*q + r (q, r)

Return type:

(int, int)

>>> cpp.find_q_r(5)
(6, 2)
cpp.bgc(n, s=None) list

Note

Works only for n=4 and n=5.

Parameters:
  • n (int) – number of bits

  • s (list) – transition sequence for n-2 bit balanced Gray code

Returns:

transition sequence for n bit balanced Gray code

Return type:

list

>>> cpp.bgc(4, s = None)
[1, 2, 1, 3, 4, 3, 1, 2, 3, 2, 4, 2, 1, 4, 3, 4]
cpp.n_bgc(n): -> list
Parameters:

n (int) – number of bits

Returns:

transition sequence for n bit balanced Gray code

Return type:

list

>>> cpp.n_bgc(6)
[1, 2, 1, 3, 4, 3, 1, 2, 3, 2, 4, 2, 1, 4, 3, 5, 3, 4, 1, 2, 4, 6, 4, 2, 1, 4, 3, 5, 3, 4, 1, 2, 4, 2, 5, 6, 3, 6, 5, 2, 5, 6, 1, 6, 5, 3, 5, 6, 4, 6, 5, 3, 5, 6, 1, 6, 5, 2, 5, 6, 1, 6, 5, 6]
cpp.computing_ab_i_odd(s_2, l, v): -> list

Note

Intrinsic function for cpp.m_length_BGC(), can not be used globally.

Parameters:
  • s_2 (list) – transition sequence for balanced Gray code with n bits

  • l (int) – number, correponds to _l_ from the method described by Lu Wang et al., 2016

  • v (int) – number, correponds to _v_ from the method described by Lu Wang et al., 2016

Returns:

[v, a_values, E_v]

Return type:

list

cpp.m_length_BGC(m, n): -> list
Parameters:
  • m (int) – required length of the code

  • n (int) – number of bits

Returns:

transition sequence for n bit balanced Gray code of length m

Return type:

list

>>> cpp.m_length_BGC(m=28, n=5)
[0, 1, 2, 3, 2, 1, 0, 4, 0, 1, 2, 3, 2, 1, 0, 1, 3, 4, 2, 4, 3, 1, 3, 4, 0, 4, 3, 4]
cpp.gc_to_address(s_2, iters, n): -> list

Tip

We do not recommend to use this function for address arrangement since the result might be imbalanced and with other features hindering the interpretation of the experiment.

Parameters:
  • s_2 (list) – transition sequence for Gray code

  • iters (int) – peptide occurrence

  • n (int) – number of pools

Returns:

address arrangement based on the produced Gray code

Return type:

list

>>> cpp.gc_to_address(cpp.m_length_BGC(m=28, n=5), 2, 5)
[[0, 4], [2, 4], [2, 3], [3, 4], [0, 3], [0, 2], [1, 3], [1, 2], [1, 4]]
cpp.union_address(address, union): -> list
Parameters:
  • address (string) – address in bit view

  • union (string) – union in bit view

Returns:

unions possible after given union and address

Return type:

list

>>> cpp.union_address('110000', '111000')
['110100', '110010', '110001']
cpp.address_union(address, union): -> list
Parameters:
  • address (string) – address in bit format

  • union (string) – union in bit format

Returns:

addresses possible after given address and union

Return type:

list

>>> cpp.address_union('011000', '111000')
['110000', '101000']
cpp.hamiltonian_path_AU(size, point, t, unions, path=None): -> list

Note

This function is recursive. It is intrinsic function for cpp.address_rearrangement_AU(), though it can work globally.

Parameters:
  • size (int) – length of the required path

  • point (string) – union or address that is added currently at this step

  • t ('a' or 'u') – type of added point (union or address)

  • unions (list) – unions used in the path

  • path (list) – addresses used in the path

Returns:

arrangement of addresses in bit format

Return type:

list

>>> cpp.hamiltonian_path_AU(size=10, point = '110000', t = 'a', unions = ['111000'])
['110000', '100100', '000110', '000011', '001001', '010001', '010010', '011000', '001100', '101000']
cpp.variance_score(bit_sums, s): -> float
Parameters:
  • bit_sums (list) – current distribution of peptides across pools

  • s (string) – union or address that is added currently at this step

Returns:

penalty for balance distortion upon this point addition to the path

Return type:

float

>>> cpp.variance_score([2, 4, 4, 3, 3, 4], '110001')
0.25
cpp.return_address_message(code, mode): -> string or list
Parameters:
  • code (list of string) – address (for example, [0, 1, 2]) or address in bit format (for example, ‘111000’)

  • mode ('a' or 'mN') – indicates whether code is address or address in bit format, if latter, than second letter (N) indicates number of pools

Returns:

corresponding address in bit format (‘111000’) or address ([0, 1, 2])

Return type:

string or list

>>> cpp.return_address_message([1, 2, 4], 'm7')
'0110100'
>>> cpp.return_address_message('0111100', 'a')
[1, 2, 3, 4]
cpp.binary_union(bin_list): -> list
Parameters:

bin_list (list) – list of addresses

Returns:

list of their unions

Return type:

list

>>> cpp.binary_union(['110000', '100001', '000101', '000110', '001010', '010010', '010100', '100100', '101000', '001001'])
['110001', '100101', '000111', '001110', '011010', '010110', '110100', '101100', '101001']
cpp.hamming_distance(s1, s2): -> int
Parameters:
  • s1 (string) – address in bit format

  • s2 (string) – address in bit format

Returns:

hamming distance between two addresses

Return type:

int

>>> cpp.hamming_distance('110000', '100001')
2
cpp.sum_bits(arr): -> list
Parameters:

arr (list) – current address arrangement in bit format

Returns:

peptide distribution across pools given this arrangement

Return type:

list

>>> cpp.sum_bits(['110001', '100101', '000111', '001110', '011010', '010110', '110100', '101100', '101001'])
[5, 4, 4, 6, 4, 4]
cpp.hamiltonian_path_A(G, size, pt, path=None): -> list

Note

This function is recursive. It is intrinsic function for cpp.address_rearrangement_A(), though it can work globally.

Parameters:
  • size (int) – graph representing peptide space

  • size – length of the required path

  • pt (string) – union or address that is added currently at this step

  • path (list) – addresses used in the path

Returns:

arrangement of addresses in bit format

Return type:

list

>>> cpp.hamiltonian_path_A(G = G, size = 10, pt = '11000', path=None)
['11000', '01100', '00101', '00011', '10010', '00110', '01010', '01001', '10001', '10100']
cpp.address_rearrangement_AU(n_pools, iters, len_lst) list, list

Note

Search for arrangement may take some time, especially with large parameters. Although, this function is faster than cpp.address_rearrangement_A(), since it considers both vertices and edges as it traverses the graph.

Parameters:
  • n_pools (int) – number of pools

  • iters (int) – peptide occurrence

  • len_lst (int) – number of peptides

Returns:

  1. list with number of peptides in each pool;

  2. list with address arrangement, uses both unions and addresses for its construction

Return type:

list, list

>>> cpp.address_rearrangement_AU(n_pools=12, iters=4, len_lst=250)
>>> b
[81, 85, 85, 85, 81, 82, 87, 81, 85, 81, 84, 83]
>>> lines
[[0, 1, 2, 3],[0, 1, 3, 6],[0, 1, 6, 8],[1, 6, 8, 9],[6, 8, 9, 11], ... ]
cpp.address_rearrangement_A(n_pools, iters, len_lst): -> list, list

Note

Search for arrangement may take some time, especially with large parameters. This function is slower than cpp.address_rearrangement_AU(), since it considers only vertices as it traverses the graph.

Parameters:
  • n_pools (int) – number of pools

  • iters (int) – peptide occurrence

  • len_lst (int) – number of peptides

Returns:

  1. list with number of peptides in each pool;

  2. list with address arrangement, uses both unions and addresses for its construction

Return type:

list, list

>>> cpp.address_rearrangement_A(n_pools=12, iters=4, len_lst=250)
>>> b
[82, 83, 85, 85, 83, 83, 84, 81, 83, 83, 84, 84]
>>> lines
[[0, 1, 2, 3],[0, 2, 3, 7],[0, 3, 7, 11],[0, 7, 10, 11],[7, 8, 10, 11], ... ]

Peptide overlap

cpp.string_overlap(str1, str2): -> int
Parameters:
  • str1 (string) – peptide

  • str2 (string) – peptide

Returns:

overlap length between two peptides

Return type:

int

>>> cpp.string_overlap('ASDFGHJKTYUIO', 'GHJKTYUIOTYUI')
9
cpp.find_pair_with_overlap(lst, target_overlap) list
Parameters:
  • lst (list) – ordered list of peptides

  • target_overlap (int) – overlap length

Returns:

list of lists with peptides with specified overlap length.

Return type:

list

>>> cpp.find_pair_with_overlap(lst, 16)
[['FDEDDSEPVLKGVKLHY', 'DEDDSEPVLKGVKLHYT']]
cpp.how_many_peptides(lst, ep_length) Counter object, dictionary
Parameters:
  • lst (list) – ordered list of peptides

  • ep_length (int) – expected epitope length

Returns:

  1. the Counter object with the number of epitopes shared across the number of peptides;

  2. the dictionary with all possible epitopes of expected length as keys and the number of peptides where these epitopes are present as values.

Return type:

Counter object, dictionary

>>> t, r = cpp.how_many_peptides(lst, 8)
>>> t
Counter({1: 6, 2: 1256, 3: 4})
>>> r
{'MFVFLVLL': 1,'FVFLVLLP': 1,VFLVLLPL': 1,'FLVLLPLV': 1,'LVLLPLVS': 1,'VLLPLVSS': 2, ...,}

Pooling and simulation

cpp.bad_address_predictor(all_ns): -> list

Tip

Initially it is designed for address arrangement produced by cpp.gc_to_address(). But keep in mind that produced arrangement might be imbalanced.

Parameters:

all_ns (list) – address arrangement

Returns:

address arrangement without addresses with the same unions. The function searches for three consecutive addresses with the same union and removes the middle one.

Return type:

list

>>> cpp.bad_address_predictor([[0, 1, 2, 3], [0, 1, 2, 4], [0, 1, 2, 5], [0, 1, 2, 6], [0, 1, 3, 6], [0, 1, 3, 5], [0, 1, 3, 4]])
[[0, 1, 2, 3], [0, 1, 2, 4], [0, 1, 2, 5], [0, 1, 2, 6], [0, 1, 3, 6], [0, 1, 3, 5], [0, 1, 3, 4]]
cpp.pooling(lst, addresses, n_pools) dictionary, dictionary
Parameters:
  • lst (list) – ordered list with peptides

  • addresses (list) – produced address arrangement

  • n_pools (int) – number of pools

Returns:

  1. pools – dictionary with keys as pools indices and values as peptides that should be added to this pools;

  2. peptide address – dictionary with peptides as keys and corresponding addresses as values.

Return type:

dictionary, dictionary

>>> pools, peptide_address = cpp.pooling(lst=lst, addresses=lines, n_pools=12)
>>> pools
{0: ['MFVFLVLLPLVSSQCVN','VLLPLVSSQCVNLTTRT',VSSQCVNLTTRTQLPPA', ...], 1: ['MFVFLVLLPLVSSQCVN','VLLPLVSSQCVNLTTRT','TQDLFLPFFSNVTWFHA', ...], ... }
>>> peptide_address
{'MFVFLVLLPLVSSQCVN': [0, 1, 2, 3], 'VLLPLVSSQCVNLTTRT': [0, 1, 2, 10], ... }
cpp.pools_activation(pools, epitope): -> list
Parameters:
  • pools (dictionary) – pools, produced by cpp.pooling()

  • epitope (string) – epitope present in one or several tested peptides

Returns:

pool indices where the epitope is present

Return type:

list

>>> cpp.pools_activation(pools, 'LGVYYHKN')
[0, 3, 8, 9, 11]
cpp.epitope_pools_activation(peptide_address, lst, ep_length): -> dictionary
Parameters:
  • peptide_address (dictionary) – peptide addresses, produced by cpp.pooling()

  • lst (list) – ordered list of peptides

  • ep_length (ep) – expected epitope length

Returns:

activated pools for every possible epitope of expected length from entered peptides

Return type:

dictionary

>>> cpp.epitope_pools_activation(peptide_address, lst, 8)
{'[0, 1, 2, 3]': ['MFVFLVLL', 'FVFLVLLP', 'VFLVLLPL', 'FLVLLPLV', 'LVLLPLVS'], '[0, 1, 2, 3, 9]': ['VLLPLVSS', 'LLPLVSSQ', 'LPLVSSQC', 'PLVSSQCV', 'LVSSQCVN'], '[0, 1, 3, 9, 11]': ['VSSQCVNL', 'SSQCVNLT', ...], ... }
cpp.peptide_search(lst, act_profile, act_pools, iters, n_pools, regime): -> list, list
Parameters:
  • lst (list) – ordered list of peptides

  • act_profile (dictionary) – activated pools for every possible epitope of expected length from entered peptides, produced by cpp.epitope_pools_activation()

  • act_pools (list) – activated pools

  • iters (int) – peptide occurrence

  • n_pools (int) – number of pools

  • regime ("with dropouts" or "without dropouts") – regime of simulation, with or without drop-outs

Returns:

possible peptides and possible epitopes given such activated pools

Return type:

list, list

>>> cpp.peptide_search(lst, act_profile, [0, 3, 8, 9, 11], 4, 12, 'without dropouts')
(['CNDPFLGVYYHKNNKSW', 'LGVYYHKNNKSWMESEF'], ['LGVYYHKN', 'GVYYHKNN', 'VYYHKNNK', 'YYHKNNKS', 'YHKNNKSW'])
>>> cpp.peptide_search(lst, act_profile, [0, 3, 8, 11], iters, n_pools, 'with dropouts')
(['CNDPFLGVYYHKNNKSW', 'LLKYNENGTITDAVDCA', 'LGVYYHKNNKSWMESEF', 'QPRTFLLKYNENGTITD'], ['YNENGTIT', 'LKYNENGT', 'YHKNNKSW', 'KYNENGTI', 'YYHKNNKS', 'LGVYYHKN', 'VYYHKNNK', 'NENGTITD', 'LLKYNENG', 'GVYYHKNN'])
cpp.run_experiment(lst, peptide_address, ep_length, pools, iters, n_pools, regime) pandas DataFrame

Note

Simulation may take several minutes, especially upon “with drop-outs” regime.

Parameters:
  • lst (list) – ordered list with peptides

  • peptide_address (dictionary) – peptides addresses produced by pooling

  • ep_length (int) – expected epitope length

  • pools (dictionary) – pools produced by pooling

  • iters (int) – peptide occurrence

  • n_pools (int) – number of pools

  • regime (“with dropouts” or “without dropouts”) – regime of simulation, with or without drop-outs

Returns:

  1. pools – dictionary with keys as pools indices and values as peptides that should be added to this pools;

  2. peptide address – dictionary with peptides as keys and corresponding addresses as values.

Return type:

dictionary, dictionary

>>> df = cpp.run_experiment(lst=lst, peptide_address=peptide_address, ep_length=8, pools=pools, iters=iters, n_pools=n_pools, regime='without dropouts')

3D models

cpp.stl_generator(rows, cols, length, width, thickness, hole_radius, x_offset, y_offset, well_spacing, coordinates): -> Mesh object
Parameters:
  • rows (int) – int

  • cols (int) – number of columns in your plate with peptides

  • length (float) – length of the plate in mm

  • width (float) – width of the plate in mm

  • thickness (float) – desired thickness of the punch card, in mm

  • hole_radius (float) – the radius of the holes, in mm, should be adjusted to fit your tip

  • x_offset (float) – the margin along the X axis for the A1 hole, in mm

  • y_offset (float) – the margin along the Y axis for the A1 hole, in mm

  • well_spacing (float) – the distance between wells, in mm

  • coordinates (list) – coordinates of holes, in tuples in list

Returns:

punch cards with holes based in entered coordinates

Return type:

Mesh object

>>> cpp.stl_generator(rows = 16, cols = 24, length = 122.10, width = 79.97, thickness = 1.5, hole_radius = 4.0 / 2, x_offset = 9.05, y_offset = 6.20, well_spacing = 4.5, [(1, 1), (2, 2), (1, 2)])
Mesh object
cpp.pools_stl(peptides_table, pools, rows=16, cols=24, length=122.10, width=79.97, thickness=1.5, hole_radius=4.0 / 2, x_offset=9.05, y_offset=6.20, well_spacing=4.5) dictionary

Note

Rendering of 3D models will take some time.

Parameters:
  • peptides_table (pandas DataFrame) – table representing the arrangement of peptides in a plate, is not produced by any function in the package

  • pools (pandas DataFrame) – table with a pooling scheme, where one row represents each pool, pool index is the index column, and a string with all peptides added to this pool separated by “;” is “Peptides” column.

  • rows (int) – int

  • cols (int) – number of columns in your plate with peptides

  • length (float) – length of the plate in mm

  • width (float) – width of the plate in mm

  • thickness (float) – desired thickness of the punch card, in mm

  • hole_radius (float) – the radius of the holes, in mm, should be adjusted to fit your tip

  • x_offset (float) – the margin along the X axis for the A1 hole, in mm

  • y_offset (float) – the margin along the Y axis for the A1 hole, in mm

  • well_spacing (float) – the distance between wells, in mm

Returns:

dictionary with Mesh objects, where key is pool index, and value is a Mesh object of a corresponding punch card.

Return type:

dictionary

>>> meshes_list = cpp.pools_stl(peptides_table, pools, rows = 16, cols = 24, length = 122.10, width = 79.97, thickness = 1.5, hole_radius = 2.0, x_offset = 9.05, y_offset = 6.20, well_spacing = 4.5)

Generated STL file you can check using OpenSCAD:

_images/pools_stl.png
cpp.zip_meshes_export(meshes_list) None
Parameters:

meshes_list (dictionary) – dictionary with Mesh objects, generated by cpp.pools_stl()

Returns:

export Mesh objects as STL files in .zip archive.

Return type:

None

>>> cpp.zip_meshes_export(meshes_list)
cpp.zip_meshes(meshes_list): -> BytesIO object
Parameters:

meshes_list (dictionary) – dictionary with Mesh objects, generated by cpp.pools_stl()

Returns:

zip archive with generated STL files in BytesIO format (suitable for emails)

Return type:

BytesIO

>>> cpp.zip_meshes(meshes_list)
<_io.BytesIO at 0x1d42a1440>