A computational method to predict the probability of lncRNA localizing to the cytoplasm

CytoLNCpred is a tool developed by Raghva-Lab in 2024. It is designed to predict the probability of lncRNA localizing to the cytoplasm. It utilizes a large language model - DNABERT-2 to make predictions. CytoLNCpred is also available as web-server at https://webs.iiitd.edu.in/raghava/cytolncpred. Please read/cite the content about the CytoLNCpred for complete information including algorithm behind the approach.

PIP version is also available for easy installation and usage of this tool. The following command is required to install the package

pip install cytolncpred

To know about the available option for the pip package, type the following command:

cytolncpred -h

The Standalone version of CytoLNCpred is written in python3 and following libraries are necessary for the successful run:

• transformers==4.29
• argparse
• biopython
• torch
• numpy
• pandas

To know about the available option for the stanadlone, type the following command:

python cytolncpred.py -h

To run the example, type the following command:

python cytolncpred.py -i example_input.fa

This will predict the probability whether a submitted sequence will localize to the cytoplasm or nucleus. It will use other parameters by default. It will save the output in "outfile.csv" in CSV (comma separated variables).

usage: cytolncpred.py [-h] -i INPUT [-o OUTPUT] [-m MODEL] [-t THRESHOLD] [-w WORKDIR]
                      [-d {1,2,3}]

Provide the following inputs for a successful run

optional arguments:
  -h, --help            show this help message and exit
  -i INPUT, --input INPUT
                        Input: nucleotide sequence in FASTA format
  -o OUTPUT, --output OUTPUT
                        Output: File for saving results; by default outfile.csv
  -m MODEL, --model MODEL
                        Model path: Folder containing the fine-tuned DNABERT-2 model; by
                        default - dnabert2_10k
  -t THRESHOLD, --threshold THRESHOLD
                        Threshold: Value between 0 to 1; by default 0.5
  -w WORKDIR, --workdir WORKDIR
                        Working directory: Directory where all intermediate and final files
                        will be created; by default .
  -d {1,2,3}, --display {1,2,3}
                        Display: 1:Cytoplasm-localized, 2: Nucleus-localized, 3: All; by
                        default 3

Input File: It allow users to provide input in the FASTA format.

Output File: Program will save the results in the CSV format, in case user does not provide output file name, it will be stored in "outfile.csv".

Threshold: User should provide threshold between 0 and 1, by default its 0.5.

Working Directory: Directory where intermediate files as well as final results will be saved

Display type: This option allow users to fetch either only lncRNA localizing to Cytoplasm by choosing option 1 or only lncRNA localizing to Nucleus by choosing option 2 or prediction for all lncRNAs by choosing option 2.

It contains the following files, brief description of these files given below

INSTALLATION : Installations instructions

LICENSE : License information

README.md : This file provide information about this package

dnabert2_10k.zip (not provided) : This zipped file contains the fine-tuned DNABERT-2 model. It can be downloaded from the link. This file should be present in the working directory.

cytolncpred.py : Main python program

example.fasta : Example file contain lncRNA sequences in FASTA format

sample_output.csv : Example output file for the program