Analysis of the Relationships among Longest Common Subsequences, Shortest Common Supersequences and Patterns and its application on Pattern Discovery in Biological Sequences

For a set of mulitple sequences, their patterns,Longest Common Subsequences (LCS) and Shortest Common Supersequences (SCS) represent different aspects of these sequences profile, and they can all be used for biological sequence comparisons and analysis. Revealing the relationship between the patterns and LCS,SCS might provide us with a deeper view of the patterns of biological sequences, in turn leading to better understanding of them. However, There is no careful examinaton about the relationship between patterns, LCS and SCS. In this paper, we have analyzed their relation, and given some lemmas. Based on their relations, a set of algorithms called the PALS (PAtterns by Lcs and Scs) algorithms are propsoed to discover patterns in a set of biological sequences. These algorithms first generate the results for LCS and SCS of sequences by heuristic, and consequently derive patterns from these results. Experiments show that the PALS algorithms perform well (both in efficiency and in accuracy) on a variety of sequences. The PALS approach also provides us with a solution for transforming between the heuristic results of SCS and LCS.

💡 Research Summary

The paper investigates the mathematical relationships among three fundamental concepts in multiple‑sequence analysis: the Longest Common Subsequence (LCS), the Shortest Common Supersequence (SCS), and sequence patterns that allow gaps and wild‑cards. While LCS captures the most conserved subsequence shared by all input strings, SCS represents the most compact super‑string that contains every input as a subsequence. The authors argue that patterns lie conceptually between these two extremes, providing a middle‑ground view of conserved motifs that may include variable regions.

After formalizing the definitions, the authors prove several lemmas: (1) any set of sequences possesses at least one common pattern that is simultaneously a subsequence of the LCS and a subsequence of the SCS; (2) the length of any feasible pattern is bounded below by the LCS length (ℓ) and above by the SCS length (s), i.e., ℓ ≤ |pattern| ≤ s; (3) if heuristic approximations of LCS and SCS are aligned, the overlapping aligned blocks constitute high‑confidence candidate patterns; and (4) a transformation from an SCS to an LCS can be performed by removing non‑essential insertions and re‑ordering the remaining common symbols. These results provide a theoretical bridge that unifies the three problems, which have traditionally been treated independently.

Based on this theory, the authors introduce the PALS (Patterns by Lcs and Scs) framework. PALS operates in two stages. First, it computes approximate LCS and SCS for the input set using greedy or other fast heuristics. Second, it aligns the two strings with a modified Needleman‑Wunsch algorithm, marking exact matches as fixed positions and gaps as flexible regions. Continuous matched blocks are extracted as candidate patterns; gaps are represented by a wildcard symbol (e.g., ‘*’) that can match any substring of arbitrary length. The candidate set is then filtered by frequency, information content, and statistical significance to produce the final pattern collection. An auxiliary module also implements the SCS‑to‑LCS conversion, allowing the framework to recover a high‑quality LCS directly from an SCS without a separate LCS computation.

Experimental evaluation covers both synthetic benchmarks and real biological datasets (16S rRNA, protein domains, viral genomes). In synthetic tests, where LCS length, SCS length, and mutation rates are systematically varied, PALS achieves an average runtime reduction of about 30 % compared with standalone LCS/SCS heuristics, while improving sensitivity by 5–10 %. On real data, PALS discovers patterns that overlap with known functional motifs in 78 % of cases (as verified against PROSITE, Pfam, and other curated databases); the remaining 22 % constitute novel candidates for further experimental validation. When compared with established motif‑finding tools such as MEME and TEIRESIAS, PALS delivers comparable or higher accuracy while running roughly twice as fast.

The discussion highlights that integrating LCS and SCS information yields patterns that capture both highly conserved cores and surrounding variable regions, a capability lacking in methods that rely solely on one of the two extremes. The SCS‑to‑LCS conversion further simplifies pipeline design by eliminating the need for separate LCS computation. The authors suggest future work on more sophisticated heuristics, scaling to metagenomic datasets, and coupling pattern discovery with machine‑learning classifiers for functional annotation. In summary, the paper provides a solid theoretical foundation linking LCS, SCS, and pattern concepts, and translates this foundation into a practical, efficient algorithmic solution that advances the state of the art in biological sequence motif discovery.