JAli

JAli applies a jumping-extension of the Smith-Waterman algorithm to align candidate sequences to multiple sequence alignments, leveraging vertical and horizontal information to classify proteins and detect remote homologs.

Key Features:

• Jumping alignment algorithm: Aligns each candidate sequence against a reference sequence within a multiple sequence alignment while allowing the reference sequence to change and applying penalties for such jumps.
• Balanced use of alignment information: Integrates both vertical (column-wise) and horizontal (sequence-wise) information from multiple alignments, contrasting with profile-based techniques and family pairwise searches.
• Improved database-search sensitivity: Demonstrates higher success in identifying candidate sequences belonging to specific superfamilies compared to profiles, profile hidden Markov models, and family pairwise search methods.

Scientific Applications:

• Protein classification: Determines membership of proteins within given superfamilies to support functional and evolutionary interpretation.
• Detection of remote homologs: Identifies distant evolutionary relationships between proteins to aid functional annotation and discovery of novel connections.

Methodology:

Each candidate sequence is tested against a multiple alignment of known superfamily members; the jumping alignment computes local alignments by allowing dynamic shifts to different reference sequences within the alignment and penalizes those jumps; performance is evaluated using median false positive counts (med-FP-counts) on subsets of the SCOP database and compared to profiles, profile hidden Markov models, and family pairwise search methods.