Bioinformatics of Protein Subcellular Localization

Talk Outline

Schematic of Joe Cell

Motivation

Why Predict?

Why not use sequence similarity alone?

Protein Subcellular Localization (aka Protein Sorting) in Eukaryotes

Organelles and Function

Sorting Signals Often Compared to Postal Address

Protein Trafficking Pathways

Representative N-terminal Sorting Signals

N-terminal signals cont.

N-terminal signals largely independent of carrier protein

Sequence Logo for Eukaryotic Signal Peptide

C-terminal Sorting Signals

Internal Sorting Signals

Common to Most Predictive Programs

Prediction Work by Group

"*P" Group

“*P” Approach

Non-linear discrimination

Ways to Achieve Non-linear Discrimination

TargetP Schematic

*Loc Group

“*Loc”Group Approach

Nuclear Import Occurs in Folded State

PSORT Group

PSORT Group Approach

Expert System Approach

Problems with Expert System Approach

Probabilistic Inference

Nair & Rost LOCtree Structure

Black Box Group

Black Box Approach

Black Box Approach(2)

PSORT-B group

Two kinds of Correlation

Two Kinds of Correlation

(non)-Causal Correlations Example

(non)-Causal Correlations Example Revisited

Summary of Causality Discussion

WoLF PSORT

Protein Localization Prediction with WoLF PSORT

WoLF PSORT Dataset

WoLF PSORT Classification Method

Localization Sites

kNN Classifier

Feature Selection

Forward Feature Selection

WoLF PSORT Schematic

WoLF PSORT User Experience

Treacher Collins syndrome: severe dominant inherited disorder. Cranio-facial defects. Mechanism speculated to be haplo-insufficiency.

wolfpsort.org

List of Nearest Neighbors

Feature Level Details

Where are the signals?

Jackknife Test Accuracy

Combined with BLAST

Dual Localization Prediction