Direct whole-plastome alignment systematically inflates long-branch artifacts in chloroplast phylogenomics: a 468-paper audit and three-method reanalysis
A comprehensive audit of 468 chloroplast phylogenomics papers reveals that direct whole-plastome multiple sequence alignment systematically inflates long-branch artifacts due to unaddressed structural complexities like inverted repeats and gene rearrangements, necessitating the adoption of structural-aware alignment strategies to ensure phylogenetic accuracy.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Plants carry a small, self-contained library of genetic instructions inside their chloroplasts, the green organelles that turn sunlight into food. For decades, scientists have used these libraries to trace the family tree of the plant kingdom, figuring out which species are closely related and which are distant cousins. The standard way to do this involves taking the genetic code from specific, stable parts of the library, lining them up side-by-side with the same parts from other plants, and looking for differences to map out their history. This method works well because it treats the genetic code like a set of distinct chapters, ensuring that the comparison is fair and accurate. However, the full genetic library has a complex structure, with sections that can flip, swap, or change length between different species. When researchers try to line up the entire library at once, rather than just the stable chapters, they risk forcing mismatched sections together, creating a false picture of how the plants are related.
A recent study by Zhi-Jie Xu at Donghua University investigated how often this shortcut of lining up entire genetic libraries causes errors. The researcher examined 468 scientific papers published between late 2023 and late 2026 that used chloroplast genomes to build plant family trees. The goal was to see if the common practice of aligning whole genomes was systematically creating misleading results, specifically making some plant branches look artificially long and distant from their relatives. By re-running the analysis of the most suspicious cases using three different, more careful methods, the study found that the shortcut was indeed causing trouble. In about 7.7 percent of the papers reviewed, the long, strange branches that appeared in the original published trees vanished or became much shorter when the data was analyzed with more rigorous techniques. This suggests that for a significant number of recent studies, the family tree was distorted not by the biology of the plants, but by the way the genetic data was lined up.
The problem stems from the physical nature of the chloroplast genome. It is organized into four parts, with two large sections that are mirror images of each other and a small section that can flip its orientation. When scientists assemble the genome from short DNA fragments, the computer often cannot tell which way the small section is facing. If a researcher then tries to align the entire genome of one plant against another without checking this orientation, the computer might force the flipped section to match the non-flipped one. This creates a massive amount of artificial gaps and mismatches in the alignment. The computer interprets these forced mismatches as a huge amount of evolutionary change, which in turn makes the branch leading to that plant look much longer than it really is. This phenomenon, known as long-branch attraction, can trick the computer into placing the plant in the wrong spot on the family tree, grouping it with unrelated species simply because they both share this alignment error.
To test how widespread this issue was, the researcher audited papers from a journal that publishes a high volume of plant genome studies. From the 468 papers, 53 were flagged as having suspiciously long branches or strange tree shapes. Each of these candidates was then re-analyzed using three independent strategies. The first strategy mimicked the common workflow used in the original papers, aligning the whole genome directly. The second strategy broke the genome into small, locally matching blocks that respected the natural structure of the DNA, ignoring the rearranged parts. The third strategy focused only on single-copy genes that are present in all the plants being studied, avoiding the messy, variable regions entirely. The results were clear: in 36 of the 53 cases, the original long branches were reproduced by the first method but disappeared or shortened significantly when the second and third methods were used. This confirmed that the original trees were likely artifacts of the alignment process rather than true biological signals.
The study also dug deeper to find the physical cause of these errors. By comparing the DNA sequences of the problematic plants against normal ones, the researcher found that the vast majority of the errors were linked to specific structural issues. In 17 cases, the small section of the genome had been flipped in the wrong direction, a mistake that the alignment software could not handle. In another 16 cases, the genome had undergone more complex rearrangements that scrambled the order of the genes. Only two cases were due to the expansion of the mirror-image sections. This structural diagnosis confirmed that the errors were not random; they were tied directly to how the DNA was organized and how the researchers had chosen to align it. The study noted that in the year 2023, the rate of these errors was notably higher, around 20 percent, but dropped to about 6.7 percent in the following years, suggesting that the community may be slowly becoming more aware of the issue, though it remains a persistent problem.
A critical finding of the audit was the lack of transparency in how these studies were reported. While all 36 papers with confirmed errors had supplementary files attached, none of them included the actual computer-readable data files, such as the alignment matrix or the specific rules used to build the tree. Without these files, other scientists cannot check the work or repeat the analysis to verify the results. The researcher argued that this is a major barrier to scientific progress. The technology to share these files already exists, and requiring them would not be difficult. By making the raw data and the alignment rules available, journals could ensure that the family trees they publish are robust and reproducible. The study concluded that while direct whole-genome alignment is a tempting shortcut, it is not a neutral one; it systematically inflates the appearance of evolutionary distance and can lead to incorrect conclusions about plant relationships. To avoid these pitfalls, researchers are encouraged to extract specific genes or use structural-aware tools that respect the unique architecture of the chloroplast genome, and to always check for structural flips before building a tree.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.