Systematic mapping of a ribozyme fitness landscape reveals a frustrated evolutionary network for self-aminoacylating RNA
Source files and python scripts used for the publication Mapping a comprehensive ribozyme fitness landscape reveals a frustrated evolutionary network for self-aminoacylating RNA by Abe D. Pressman, Ziwei Liu, Evan Janzen, Celia Blanco, Ulrich F. Muller, Gerald F. Joyce, Robert Pascal and Irene A. Chen.
- Python 2.7
- openpyxl package for python
- Levenshtein package for python
- numpy package for python
- scipy package for python
- argparse package for python
- heapq package for python
This tool calculates A and k*t, according to the equation A(1-Exp(-k[S]t)), for every sequence in a relevant k-Seq data set. It requires several input files: a file with sequence data/counts for a k-Seq "start" round, a file for each tested k-Seq round after selection has occurred, files describing the reaction conditions, and known or approximate normalization constants for each round based on the expected amount of DNA/RNA/protein present before and after the reaction in each sample.
To reproduce the numerical results reported in this publication, the python script kseq_tools_v01.py can be run as follows:
python kseq_tools_v01.py start_round kseq_rounds output_file normalization_list substrate_concs rounds_to_average rounds_to_error -v -p
where start_round is R5c-counts.txt, kseq_rounds is example-rounds.txt, normalization_list is example-normalization.txt, substrate_concs is example-subst-concs.txt, rounds_to_average is example-rnds-to-avg.txt and rounds_to_error is example-rnds-to-err.txt.
The code requires "counts" files as input, consisting of three lines of metadata followed by one line per unique sequence in the pool in the following format: sequences in the first column and counts (an integer number) in the second column. Such files are produced by our Galaxy tools, currently available at the Chen Lab website. The files corresponding to the start round (R5c-counts.txt) and each tested k-Seq round after selection (called in kseq_rounds) can be found in Dryad. Every other input file can be found in the folder kseq_tools.
For more information on usage, see the detailed readme file or run python kseq_tools_v01.py -h in the terminal.
The shortest pathway between two sequences, along a fitness landscape, can be found efficiently using the A* algorithm. The algorithm iterates over a single round's sequence population as a graph, with each sequence present as its own node and the edit distances between sequences as edges with distance. The python script peak_pather_v01.py finds the N best pathways between two sequences, ranked by: 1) shortest total path length, 2) smallest maximum step size, 3) smallest average step size, 4) largest minimum sequence count, using a sequence counts file as the reference map. Among equal-length pathways, the ones with the highest number of total steps are prioritized.
The pathways described in this publication are computed as follows:
python peak_pather_v01.py input_file output_file start_sequence end_sequence --min_count [variable] --max_step [variable] -n [variable] --max_length [variable] -p
where input_file is R5c-counts.txt. The code requires a "counts" file as input, consisting of three lines of metadata followed by one line per unique sequence in the pool in the following format: sequences in the first column and counts (an integer number) in the second column. Such files are produced by our Galaxy tools, currently available at the Chen Lab website.
The file corresponding to the round used to produce the pathways reported in the publication (R5c-counts.txt) can be found in Dryad.
Minimum sequence count (min_count) was set at 3 for the highly-populated pathways between Motif 1A and 1B, and set at 2 for all other pathways. For all pairs of sequence endpoints investigated, maximum step size (max_step) was set at 1, then incremented by 1 until 5 pathways were found. The script was then run again with min_step increased 1 further, to generate 5 additional pathways with larger step tolerance.
The number of best pathways to be generated with each run (n) was set to 5. The maximum length of any path searched before the path is abandoned (max_length) was set to 35.
For more information on usage, see the detailed readme file or run python peak_pather_v01.py -h in the terminal.
The ruggedness of a ribozyme family can be measured using the fitness correlation ,
which is the average correlation of activity effects of single mutations in d-mutant neighbors
(where d is the Levenshtein edit distance, i.e., the number of substitutions, insertions or
deletions between two related sequences).
The correlation of fitness effects,
,
is calculated following Ferretti et al., 2016.
To reproduce the numerical results reported in the publication, the python script ActivityCorrelationGamma.py can be run as:
python ActivityCorrelationGamma.py input_file sheet_name correlation_distance
where input_file corresponds to the excel file ActivityObservedData.xlsx and sheet_name corresponds to either Family_2.1, Family_1A.1, Family_1B.1, Family_1B.2 or Family_1A.2. Correlation_distance is an integer number {0..4}.
The excel source file ActivityObservedData.xlsx can be found in the folder Gamma_correlation. Each spreadsheet in the excel file corresponds to a different family and follows the same format: column A) sequences, column B) calculated activity (including values under the baseline activity), column C) calculated activity if above the baseline activity (or baseline activity if below), and column D) logarithm of the values in column C).
- Abe Pressman - k-Seq Analysis - Evolutionary Pathways
- Celia Blanco - Correlation of Fitness Effects