MGEScan-LTR

MGEScan-LTR identifies long terminal repeat (LTR) retrotransposons de novo in genomic sequences using sequence pattern matching and protein-domain evidence to detect intact elements.

Key Features:

• De Novo Identification: Performs library-independent de novo identification of LTR retrotransposons without relying on existing databases.
• Approximate String Matching: Employs approximate string matching to locate candidate LTR sequences.
• Protein Domain Analysis: Confirms retrotransposon coding regions by detecting protein domains and evaluating them with profile hidden Markov models (profile HMMs).
• GHMM-Inspired State Modeling and Linker Classification: Implements generalized hidden Markov model (GHMM)-inspired states for protein domains and inter-domain linker regions and applies Gaussian Bayes classifiers for linker evaluation.
• Genome-Wide Application: Applicable to genome-wide analyses across diverse eukaryotic genomes.

Scientific Applications:

• Genome Annotation: Supports comprehensive annotation of eukaryotic genomes by identifying LTR retrotransposons across entire genomes.
• Evolutionary Studies: Enables evolutionary and comparative analyses by detecting and classifying LTR elements to study their distribution and diversity across species.
• Novel Element Discovery and Comparative Detection: Demonstrated recovery of all known full-length elements in Drosophila melanogaster (classified into clades CR1, I, Jockey, LOA, and R1) and detection of more elements in Daphnia pulex than RepeatMasker using the RepBase Update library, and identification of novel elements in other eukaryotic genomes.

Methodology:

Uses approximate string matching to locate LTR candidates followed by protein domain analysis; modeling is GHMM-inspired with states representing protein domains and inter-domain linkers, protein domains evaluated by profile HMMs, and linker regions classified by Gaussian Bayes classifiers.