About GoLizard

GoLizard is a bioinformatics analysis pipeline designed to pre-annotate unknown or mutated protein sequences. It leverages the Gene Ontology (GO) framework and integrates multiple inference strategies, including domain analysis, sequence similarity, structural homology, and motif detection.

---

Introduction

The increasing discovery of novel and mutated proteins has created a critical gap in functional annotation. Traditional approaches often rely solely on sequence or structure-based inference, which limits their effectiveness. GoLizard addresses this challenge by combining multiple complementary strategies to enhance GO-based functional interpretation.

Analysis Modes

GoLizard offers two distinct analysis modes:

• Functional Analysis: Predicts per-sequence functionality, generating BP (Biological Process), MF (Molecular Function), and CC (Cellular Component) graphs for each protein sequence.
• Similarity Analysis: Clusters multiple sequences based on semantic similarity (GO-term similarity) using GOGOsim (Zhao & Wang, 2018) with a weighted BMA algorithm. Produces dendrograms showing sequence relationships and ranked GO terms with evidence scores.

Core Tools and Methodology

• InterProScan: Identifies annotated domains from multiple member databases.
• BLAST: Identifies sequence similarity to known proteins for GO transfer.
• ESMfold: Predicts protein structure via BioLM.ai
• FoldSeek: Detects structural homology for function prediction.
• ELM (Eukaryotic Linear Motif): Searches for functional motifs in disordered regions.
• GOGOsim: Algorithm to compute semantic similarity between proteins (Zhao & Wang, 2018).

Input and Output

Input:

• FASTA file containing one or more protein sequences.
• Similarity analysis requires multiple sequences for clustering.
• Optional: PDB structure files for structural analysis.

Output:

• Functional Analysis: BP, MF, CC graphs for each sequence showing predicted GO terms and their relationships.
• Similarity Analysis:
  • Ranked GO terms with associated evidence scores for each sequence.
  • Dendrograms of sequence clusters based on semantic similarity.

Applications

• Pre-annotation of unknown protein sequences.
• Functional annotation of proteins in non-model organisms.
• Mutational impact analysis in proteomics studies.
• Characterization of hypothetical or novel proteins.
• Predictive insights into pathogen protein function.

Contact

For questions or support, please contact:

• marek.pospisil@uochb.cas.cz
• jiri.vondrasek@uochb.cas.cz