yyh
/
FScanpy-package
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

修复图片显示异常

This commit is contained in:
yyh 2025-03-18 11:32:06 +08:00
parent 6e37c64608
commit a3c8aa99c7
1 changed files with 2 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
tutorial/tutorial.md
View File

@ -2,11 +2,11 @@
FScanpy is a Python package designed to predict Programmed Ribosomal Frameshifting (PRF) sites in DNA sequences. It integrates advanced machine learning models, including Gradient Boosting and BiLSTM-CNN, to provide accurate predictions. This tool is essential for understanding gene expression regulation in various organisms, including eukaryotes and viruses, and offers a robust solution for PRF prediction challenges.
## Introduction
![FScanpy structure](/tutorial/image/structure.jpeg)
![FScanpy structure](/image/structure.jpeg)
FScanpy is a Python package dedicated to predicting Programmed Ribosomal Frameshifting (PRF) sites in DNA sequences. It integrates machine learning models (Gradient Boosting and BiLSTM-CNN) along with the FScanR package to furnish precise PRF predictions. Users are capable of employing three types of data as input: the entire cDNA/mRNA sequence that requires prediction, the nucleotide sequence in the vicinity of the suspected frameshift site, and the peptide library blastx results of the species or related species. It anticipates the input sequence to be in the + strand and can be integrated with FScanR to augment the accuracy.
![Machine learning models](/tutorial/image/ML.png)
![Machine learning models](/image/ML.png)
For the prediction of the entire sequence, FScanpy adopts a sliding window approach to scan the entire sequence and predict the PRF sites. For regional prediction, it is based on the 33-bp and 399-bp sequences in the 0 reading frame around the suspected frameshift site. Initially, the Gradient Boosting model will predict the potential PRF sites within the scanning window. If the predicted probability exceeds the threshold, the BiLSTM-CNN model will predict the PRF sites in the 399bp sequence.Then,VotingClassifier will combine the two models to make the final prediction.
For PRF detection from BLASTX output, FScanpy identifies potential PRF sites from BLASTX alignment results, acquires the two hits of the same query sequence, and then utilizes frameDist_cutoff, mismatch_cutoff, and evalue_cutoff to filter the hits. Finally, it employs [FScanR](https://github.com/seanchen607/FScanR.git) to identify the PRF sites.