TCRnumba 0.2.6

some description

Installation

To install on Linux type into the commandline:

pip install TCRnumba

It requires CUDA and is not yet tested for MacOS or Windows.

Usage

In Python you can import

import TCRnumba

Then, test the script, by running the additional commands

import TCRnumba.graph_numba as gn
import TCRnumba.convert_pure as cp
adjacency_sparse = gn.adjacency_matrix(['ABC', 'DEF', 'EFG', 'ABC', 'ABD', 'EEE', 'EEF', 'EFF', 'GGE', 'GAS'], N_part=2, len_xy=5, direct_output=True)
cp.convert_pure(adjacency_sparse, 2, 5)

Use SONIA to create files

sonia-generate --humanTRB -n 1000000 --pre -o pre_example.txt

Functions

convert_pure(sparse, single_sidelength, len_x)

Takes list of single digit indices and convert into a dense matrix.

dense_matrix(data, len_xy)

Takes list of two digit indices and convert that into a dense matrix.

total_idx(el, N_part, i, j, len_x, len_y)

Converts indices from single-number indices to two-number indices. "N_part" is the number of strings in each block. i and j are int numbers, referring to the index of the block. len_x and len_y are int numbers denoting the number of blocks in height and length.

adjacency_matrix(seq, name="sparse.txt", idx_max=0, name_params="data/sparse_params.txt", N_part=None, len_xy=None, direct_output=False)

Inputs a list of strings "seq" and calculate the distance matrix. Save under the filename "name". // define the other params of all the presented functions

robustness distribution

(a) The relative size of the largest connected component (LCC) as a function of $1-p$, where $p$ is the fraction of removed nodes. The TCR network has \NSeqRobustness nodes with lmax = 1. The TCR network does not exhibit a
percolation phase transition. Moreover, the system is robust against small node removal. (b) The corresponding degree distribution of the same network is plotted for (0.01, 0.49, 0.69, 0.89) with darker green indicating a smaller p. With increasing p, the curve shifts downwards and becomes steeper as the number of nodes decreases.