Rethinking the 98% DNA Claim

When the genome becomes a jigsaw puzzle

Think of a 3 billion-piece jigsaw puzzle with no picture on the box, where the pieces come in snippets of 600 to 1,200 letters at a time. That's not just a metaphor: it's how early DNA sequencing actually looked. Sanger-style sequencing—an older but metadata-rich method—typically produced pieces about 700 to 1,200 bases long. Scientists had to assemble billions of those pieces into coherent chromosomes.

Early projects used an easier way to solve that puzzle: they used the human genome as a scaffold. As one careful observer put it, the early versions of the chimpanzee genome were assembled based on the human genome. In plain language, researchers sometimes hung chimp snippets onto a human framework to see how they fit, which effectively "humanized" early ape assemblies.

That assembly choice matters. If you assemble a puzzle by looking at a different finished picture, you bias the outcome. It's not an accusation; it is a technical reality. The way you assemble sequences shapes the similarity numbers you report.

The Heart of It: Assembly methods affect conclusions.

Behind the Words: When you choose a scaffold for a new genome, you're choosing a template that can bias alignments and apparent similarity. Older sequencing methods yielded short reads and required scaffolding; that scaffolding sometimes used the human genome so early assemblies relied implicitly on human structure.

Try This: Picture a lab bench stacked with plates and a researcher sorting tiny strips of letters like puzzle pieces. Before you accept a headline about similarity, ask that researcher, "Which picture did you use on the box?"

This matters because the size and shape of our story—the theological story about image-bearing and purpose—doesn't rise or fall on a single press headline. It rests on careful interpretation. When scientists later used longer DNA reads and built genomes de novo, they admitted earlier assemblies had been humanized, and that changes the numbers we should trust.

Why the 98% claim doesn't tell the whole story

That blunt claim—"this scientific fact isn't just wrong, it's wildly misleading"—forces us to ask whether a popular headline reflects the complexity of the data.

Many of the earliest comparisons that produced the 98–99% figure compared only regions that were already similar—protein-coding genes and other conserved regions—and discarded or ignored regions that did not fit the narrative. In other words, the comparison often looked at similarity and compared the similar parts, then reported that similarity as if it applied to the whole genome.

That is cherry-picking. It isn't always malicious; sometimes it's a consequence of what was technically feasible. But it matters that those early comparisons left out large swaths of DNA that are highly variable, repetitive, and functionally significant.

A more comprehensive approach, using longer reads and broader alignments, finds a very different number. When researchers compared single-copy long fragments and allowed gaps—using liberal alignment parameters—average identity values dropped toward the mid-80s rather than the high 90s. One careful reanalysis that used long contiguous sequences reported roughly 84–85% identity in comparable genome-to-genome alignments.

The Heart of It: Percent similarity depends on what you include.

Behind the Words: If you compare only conserved genes you will report a high similarity; include noncoding, structural, and highly variable regions and the number changes. In early genome comparisons, selecting only similar sequences produced high-percentage claims; later papers using de novo long-read assemblies reported lower identity figures when the full range of sequence features was considered.

Try This: Picture a small neighborhood photo album that shows only family portraits—if someone judged the town by those photos, they'd miss empty lots, businesses, and churches that matter to community life. Similarly, ask for a genomic album that includes the full variety of sequences before you let a statistic define a species.

Common misunderstanding: People often assume statistics are fixed truths; in genomics, percentages are derived outcomes depending on methods and choices. A single number rarely tells the full story.

How contamination can skew genetic studies—and what we do about it

If you work in a lab you learn quickly that DNA is messy and stubbornly social: human DNA sneaks into places we don't want it. In public databases, researchers have repeatedly found human sequences in bacterial genomes and other unexpected places—one notable finding was that three million bases of zebrafish chromosome three appeared entirely human.

How does that happen? DNA travels: on breath, on fingertips, through ventilation. When labs process delicate reactions, tiny amounts of stray human DNA can enter samples. In projects that mix millions of short reads, a little contamination can accumulate into an illusion of similarity.

Contamination isn't just about sloppy technique. It was sometimes invisible to early algorithms when huge datasets were queried at once. When one reanalysis found that an alignment pipeline was "kicking out DNA sequences" unexpectedly, developers fixed the algorithm for high-throughput use and the similarity numbers shifted.

The early chimp genome sequencing itself showed signs of this phenomenon. The first half of the project—sequenced with earlier protocols—was roughly 6% more similar to human than sequences produced later, suggesting contamination and procedural learning curves played real roles in how similar the published assembly looked.

The Heart of It: Contamination changes what we think we see.

Behind the Words: Laboratory contamination and metadata gaps can introduce foreign sequences that inflate perceived similarity. Studies have documented human DNA appearing in non-primate genomes; contamination is a technical reality that can skew apparent similarity unless properly detected and removed.

Try This: Imagine reading a diary where a few lines were copied from someone else and glued in. Would you treat the diary as wholly representative? When researchers find human reads in ape datasets, treat the assembly with caution and ask, "Which lines belong to which author?"

Practically, modern sequencing uses longer reads and better contamination filters, but older public data still influences textbooks and popular claims. That means we, as thoughtful Christians, should encourage churches and classrooms to prefer up-to-date, vetted analyses over repeating old slogans that may rest on mixed data.

What role do algorithms play in genetic comparisons?

Algorithms are the translators and judges of sequence data. They align fragments, allow gaps, score mismatches, and ultimately decide whether two sequences are "the same" enough to be called similar. But algorithms come with assumptions and limits, and when those limits are reached the output can mislead.

For example, an alignment tool designed for small numbers of sequences may behave very differently when asked to process millions of reads at once. One analyst discovered that the algorithm was excluding DNA sequences during high-throughput queries—sequences that should have been counted were silently ignored. After developers adjusted the software for massive datasets, alignment results changed and similarity estimates shifted.

Algorithms also have parameter choices: how many gaps to allow, how big an insertion before alignment stops, how to penalize mismatches. Loosen those parameters and you may recover alignments across larger differences; tighten them and you will report only conservative matches. That is why two competent labs can run similar data and report quite different identity percentages.

The Heart of It: Tools shape outcomes.

Behind the Words: Algorithms are not neutral—they encode design choices that influence alignments and conclusions. High-throughput sequencing introduced scale problems some alignment tools weren't designed for; once corrected, previously omitted sequences were included and percentages shifted.

Try This: Imagine scoring a race where one timer only counts laps if the runner stays in a narrow lane—and discards any lap that wanders. That timer's report would undercount a certain style of runner. Algorithms can similarly favor certain sequence patterns; ask how the timer was set.

A pastoral note: the presence of algorithmic bias doesn't mean scientists are dishonest; it means work is human and fallible. We owe it to one another to read methods, not only headlines. We can be curious and confident in God's truth while remaining humble about the limits of instrumentation.

"So God created mankind in his own image, in the image of God he created them; male and female he created them." — Genesis 1:27

Behind the Words: Genesis 1:27 is spoken in a creation-account context, affirming humanity's unique status among creatures and grounding dignity in God's intentional act.

The Heart of It: Our identity isn't decided by percentages.

Try This: Picture a Sunday table where someone brings a newspaper with a bold headline about DNA similarity. Read the piece together, ask what methods were used, and then remind one another of our created status in the image of God. Let the data inform, not replace, the theological truth already given.

Psalm 139 and human dignity

"I praise you because I am fearfully and wonderfully made; your works are wonderful, I know that full well." — Psalm 139:14

Behind the Words: David wrote Psalm 139 as an intimate reflection on God's close knowledge of the human person—formed in secret, knitted together in the womb. The original context is worshipful awe at God’s attentive creativity.

The Heart of It: Science explores mechanism; Scripture anchors meaning and worth.

Try This: When a scientific claim threatens to reduce you to a number, picture a midwife holding a newborn and whispering God's name over a life. That human scene is the church’s corrective when cultural stories attempt to shrink us.

Some read Scripture as opposed to scientific discovery; instead, Scripture provides the why, and the why informs how we interpret data without fear.

Numbers matter, but numbers alone do not tell your whole story. When the public hears "98% similar," many assume the scientific community agrees on a simple fact about identity; in truth the figure depended on choices—what was compared, how algorithms were run, and how contamination was handled. We owe it to our children and neighbors to insist on careful, transparent science and to hold fast to the theological truth that human dignity is rooted in God's image, not in a headline.

So ask questions. Read methods. Encourage your church to value scientific literacy and curiosity. When someone drops a statistic into conversation, respond with generosity and discernment: "What did they compare? Which sequences were included? Did they check for contamination?" Those are humble questions that deepen truth.

We do not fear science. We want science that is honest. That commitment honors God, serves our neighbors, and protects the fragile thing we call human identity.