gardner-binflab / tisigner_paper_2019

Code related to TIsigner manuscript and webserver

Bhandari BK, Lim CS, Remus DM, Chen A, van Dolleweerd C, Gardner PP. (2021). Analysis of 11,430 recombinant protein production experiments reveals that protein yield is tunable by synonymous codon changes of translation initiation sites. PLOS Computational Biology. 17(10), e1009461. DOI:10.1371/journal.pcbi.1009461

• This repository contains the scripts and Jupyter notebooks to reproduce the results and figures of this preprint. The source code of TIsigner webserver is available here.
• Dependencies can be installed using Anaconda3. For example, conda install -c bioconda viennarna. ViennaRNA can also be installed according to the instructions here.
• IXnos requires python2 to run.
• openen.py is a wrapper for RNAplfold using multiple processes. It is useful to calculate the opening energy of multi-fasta sequences. The output can be analysed as in Fig1_2_S1_S2.ipynb

$ python openen.py -h
usage: openen.py [-h] -s STR [-U STR/INT] [-x] [-W INT] [-u INT] [-S] [-n INT]
                 [-t INT] [-e] [-i INT] [-l INT] [-r] [-o STR] [-p INT]

RNAplfold wrapper using multiprocesses

optional arguments:
  -h, --help            show this help message and exit
  -s STR, --sequence STR
                        Sequences in fasta or csv format
  -U STR/INT, --utr STR/INT
                        Use an integer if 5UTR presence, e.g., -U 1. Use your
                        own 5UTR sequence if your plasmid backbone is not of
                        pET vector. Default = GGGGAATTGTGAGCGGATAACAATTCCCCTCT
                        AGAAATAATTTTGTTTAACTTTAAGAAGGAGATATACAT
  -x, --execute         Run RNAplfold multiprocessing
  -W INT, --winsize INT
                        Average the pair probabilities over windows of given
                        size. An RNAplfold option. Default = 210
  -u INT, --ulength INT
                        Compute the mean probability that subsegments of
                        length 1 to a given length are unpaired. An RNAplfold
                        option. Default = 210
  -S, --stack           Stack _openen dataframes to single-column dataframes,
                        concatenate them as a single pandas dataframe and
                        output it as a .pkl pickle file. Requires i and j
                        options
  -n INT, --utrlength INT
                        The length of 5UTR. Related to option -S and -e.
                        Default = 71
  -t INT, --distance INT
                        Downstream distance to start codon to include when
                        stacking. Related to option -S. Default = 100
  -e, --parse           Parsing _openen dataframes to get opening energy of
                        unpaired subsegments. Requires i and l options
  -i INT, --ipos INT    Position i centered at start codon of an input
                        sequence. Related to option -e. Default = 18
  -l INT, --length INT  Subsegment l as in _openen file. Related to option -e.
                        Default = 48
  -r, --remove          Remove _openen and .ps files
  -o STR, --output STR  Output file name for .pkl. Related to -S. Default =
                        openen
  -p INT, --processes INT
                        Number of processes to spawn. Default = half of the
                        number of CPU

© Bikash Kumar Bhandari, Chun Shem Lim, Paul P Gardner (2019-)

tisigner_paper_2019's Issues

missing rf_model

Hello,
I looked at the code in optimization.py, it seems missing rf_model.
self.rf_model = self.validate_model(rf_model).
look like you pre-trained features,
could you share the code for training?
thanks