Overview

The goal of this project is to predict signal peptide cleavage site using supervised learning.

Table of Contents

• Project Structure
• Installation
• Data
• Models
• Evaluation

Project Structure

• /projet final: Contains all the C++ code and the data for the project
• p2-cleavage.pdf contains the description of the project
• report.pdf contains the description of our solution

Installation

In order to use our codes and re-carry out our tests (if you want to check them and try them :)), there is a Makefile for the compilations and you can use "./grader x", where x is the number of the test.

Data

The projet final/data folder contrains for files containing sequences from various organisms, where the signal peptide is followed by a part of the amino acids of the mature protein:

• EUKSIG 13.red contains 1005 sequences from eucaryotes,
• GRAM+SIG 13.red contains 140 sequences from Gram-positive prokaryotes.
• GRAM-SIG 13.red contains 263 sequences from Gram-negative prokaryotes.
• SIG 13.red is a concatenation of the three files listed above.

Each entry in these files consists in exactly three lines. This is an example of a protein description: 55 2SS8 HELAN 25 ALBUMIN 8 PRECURSOR (METHIONINE-RICH 2S PROTEIN) (SFA8). MARFSIVFAAAGVLLLVAMAPVSEASTTTIITTIIEENPYGRGRTESGCYQQMEE SSSSSSSSSSSSSSSSSSSSSSSSSCMMMMMMMMMMMMMMMMMMMMMMMMMMMMM

• The first line gives a description of the protein and its content may be ignored.
• The second line gives the beginning of the primary structure of the protein, with letters encoding the amino acids. Note that the first amino acid is almost always M (methionine), as it corresponds to the start codon.
• The third line is an annotation corresponding to the localization of the cleavage site. For each amino acid on the previous line, a letter indicates whether this amino acid belongs to the signal peptide (annotated with a letter S) or to the mature protein (annotated with a letter M). The first amino acid of the mature protein is annotated with a letter C.

Models

The description of the model can be found in the report.pdf file.

Evaluation

The evaluation of the model is simply the accuracy of the prediction.