# How the DNA Model Was Built

> Choose DNA models by the question; every model leaves features out. You compare four models of DNA, follow the evidence that led to the double helix in 1953, and see how sequencing added a base-letter model and a human reference sequence. Then you read cost data and make a claim about how technology changed what biologists can study.

Course: Biology (DNA and inheritance). Level: grades 7–12. About 40 minutes.
Lesson page (link this one when citing): https://stemhop.com/lessons/discovering-dna-structure/

Builds on: DNA: Structure and Copying · What Am I Made Of?

## What you'll learn

- You will compare several models of DNA (a ladder, a double helix, a base-letter code and a space-filling model) and judge what each one shows and leaves out.
- You will explain how X-ray diffraction images, base-composition data and model building led to the double helix model in 1953, and how advances since 1990 (the Human Genome Project, which improved the Sanger sequencing method of 1977, and sequencing whole chromosomes) added to what we know about DNA’s structure and function.
- You will read data on the cost of sequencing a genome over time and make a claim about how technology changed what biologists can study.

Before you start, decide whether you agree with each statement below.

What do you think? Decide before you read.

1. Watson and Crick discovered DNA. Agree Disagree **Scientists disagree:** DNA was identified by Friedrich Miescher in 1869, in material from the nuclei of white blood cells. In 1953 Watson and Crick proposed a model of how DNA is shaped. Their model used measurements that other scientists had already made, so it was the result of many people’s work. What scientists think **Scientists disagree:** DNA was identified by Friedrich Miescher in 1869, in material from the nuclei of white blood cells. In 1953 Watson and Crick proposed a model of how DNA is shaped. Their model used measurements that other scientists had already made, so it was the result of many people’s work.
2. A scientific model is a perfect copy of the real thing. Agree Disagree **Scientists disagree:** Every model makes approximations. A flat ladder drawing of DNA removes the twist, and a string of letters gives the order of bases but no shape. What scientists think **Scientists disagree:** Every model makes approximations. A flat ladder drawing of DNA removes the twist, and a string of letters gives the order of bases but no shape.
3. The Human Genome Project read the DNA of one single person. Agree Disagree **Scientists disagree:** The first human reference sequence was a patchwork of DNA from several anonymous donors, with about 70 percent from one person. It works as a reference, a standard to compare against, rather than as a record of any one person. What scientists think **Scientists disagree:** The first human reference sequence was a patchwork of DNA from several anonymous donors, with about 70 percent from one person. It works as a reference, a standard to compare against, rather than as a record of any one person.

## Explore: four models of one molecule

In the simulation, tick what each of four views can explain (pairing, copying, information, packing) and see what it hides. First predict which of the four questions the flat ladder can answer.

*Interactive on the lesson page:* Compare four models of DNA

**Predict.** Which view would best show how DNA is packed into a cell: a flat ladder, a letter string or a chromosome view?

> **Check your prediction.** The chromosome view shows large-scale packing. It hides the individual base pairs. Then decide which model best compares one gene with a classmate’s.

Pick the one model you would keep if you could keep only one. Write what you would lose.

**The one idea:**

Choose DNA models by the question; every model leaves features out.

**Scope:**

this rule applies to every DNA model in this lesson, including a reference sequence used for comparison. It does not make all models equally accurate; evidence can rule a model out.

## What each model can answer

- model
- nucleotide
- base pairing
- double helix
- X-ray diffraction
- DNA sequencing
- genome
- Human Genome Project
- chromosome
- reference sequence

A **model** is a simplified representation that helps explain or predict something real. A city map shows streets but not the people on them.

A **nucleotide** is one building block of DNA, made of a sugar, a phosphate group and a base. DNA has four bases, written A, T, G and C. The order of the bases carries the instructions, as [the DNA structure lesson](https://stemhop.com/lessons/dna-structure-and-copying/) shows.

| Model | Good for showing | Leaves out |
| --- | --- | --- |
| Flat ladder with letters | Which bases pair, and the two backbones | The twist, and all the atoms |
| Double helix ribbon | How two strands coil around each other | Individual atoms, and often the actual base order |
| Base-letter string (ATTGC…) | The order of bases, for comparing DNA | Shape, size and the partner strand |
| Space-filling model | How much space the atoms fill | Base pairs hidden inside the molecule |

The map analogy has a limit: a map is drawn from a place you can visit. The B-form double helix is about 2 nanometers wide, far too narrow to see with your eyes or a light microscope. Its models had to be built from measurements.

## Three kinds of evidence, one model

DNA was identified in 1869, long before anyone knew its shape. By the early 1950s, the three kinds of evidence in the first film could rule models out.

| Evidence | What a model had to explain |
| --- | --- |
| Base amounts | Erwin Chargaff’s group found A close to T and G close to C in a sample. |
| X-ray diffraction | X-rays scattered from lined-up DNA fibers into spots. Their spacing pointed to a helix about 2 nanometers wide and gave repeat distances, but it was not a photograph of the bases. |
| Chemistry and model building | The charged phosphate backbones had to face the surrounding water, and the known sizes of the parts had to fit together. |

Rosalind Franklin and Raymond Gosling measured diffraction patterns, including one often called Photo 51. Their work on how DNA fibers changed as the air got wetter or drier also supported the idea that the phosphates sit on the outside, facing water.

**Predict.** A species has more G and C than A and T. Does that rule out base pairing in the double helix?

> **Check your prediction.** No. Pairing predicts only that G is close to C and A is close to T. Chargaff’s work showed that the four bases do not come in equal amounts, and that the base mix differs among species.

James Watson and Francis Crick’s paper said they were stimulated by knowing the general nature of unpublished results and ideas from Maurice Wilkins, Franklin and their coworkers at King’s College London.

Their model, published on 25 April 1953, was the **double helix**: two strands of DNA coiled around each other, running in opposite directions, with the backbones outside and bases inside. **Base pairing**, the rule that A pairs with T and G pairs with C across the two strands, explained the equal amounts of A and T. It also suggested how DNA could be copied, because each strand fixes the order of its partner.

*Interactive on the lesson page:* Tap to enlarge Base amountsX-rayslined-up DNA fiberspattern of spotsdouble helix blankX-raysblankblankdouble helix Word 

In print, Watson and Crick called their model unproved until it could be checked against more exact results, and Franklin and Gosling called a helix highly probable.

**Key idea:**

The double helix model was built to fit base-amount data, X-ray measurements and chemistry together, and later tests supported it.

Shape still did not reveal base order; the second film shows how scientists learned to read it.

**Film transcript: How DNA Became Readable**

### Film transcript: How DNA Became Readable

The double helix revealed the shape of DNA. But how can scientists read base order when shape does not show it? Move the yellow lens to order. Start with one template strand. A copying enzyme moves along it, adding matching base blocks. Ordinary blocks let the copy keep growing. The letters are still hidden in the process. Sanger's original method ran a separate reaction for each kind of base. Each reaction mixed ordinary blocks with a small amount of one kind of stop block. When a stop block joins a copy, that copy cannot grow farther. Many copies grow at once, so copies stop at many different matching positions.

Later versions give each kind of stop block its own color and use one reaction. Each tray now holds a mixture, not one planned staircase. Each copy ends with the base that stopped it. Sorting lanes can turn those lengths into readable order. Predict before the lane runs. Which copy reaches the bottom first: a shorter copy or a longer copy? Hold the picture and choose.

All the copies start moving together, but shorter copies move faster. The shortest reaches the bottom first, then longer copies follow. Reading each stop base in that order reveals the complementary sequence, one position after another. Sequencing changes the useful model. A coiled shape becomes a letter string. That string shows identity and order clearly, while hiding shape, size, and the partner strand. A genome is all the DNA in one chromosome set. A reference sequence acts like a ruler. New sequence strips align against it, making differences easier to find. The Human Genome Project used improved Sanger sequencing to build a human reference.

It joined work from several donors and still left gaps. Later projects filled more gaps and added variation. Reading also became far less costly. Plot the lesson's production-cost estimates on a logarithmic scale. The computed curve plunges, then flattens. The curve needs limits. These are production estimates, not prices paid by a person.

Later estimates use an existing reference. Earlier estimates priced building without one. Interpretation costs are excluded. Lower cost means a fixed budget can read more genomes. Biologists can compare many people and species instead of relying on one reference. The reference remains a model with limits. Shape required a structure model.

Order required a sequence model. Comparison required a reference. Choose DNA models by the question; every model leaves features out.

From the film

- Sequencing reads base order, which a shape model does not show.
- Stopped copies sorted by length, shortest first, reveal a complementary base sequence.
- A reference sequence is a comparison standard, not one person's genome.
- Lower production costs let biologists compare many genomes.

## Reading the letters since 1977

**DNA sequencing** is any method that reads the order of bases in a piece of DNA. Frederick Sanger and his colleagues Nicklen and Coulson published the method in the second film in 1977. Its core trick: a small amount of stop blocks, mixed in with ordinary ones, stops copies at many different matching positions. The 1977 method ran four separate reactions, each with a stop block for only one kind of base. So the reaction a copy came from told which base ended it. Sorted by length, shortest first, the copies reveal the base order of the new, complementary strand.

*Figure (AI-generated illustration).* The idea behind Sanger's 1977 method: copies of a strand stop at each base, and their lengths reveal the order of the bases. Image: a top row of eight lettered, coloured blocks labelled template strand (T A C G G A T C). Below it a staircase of eight copies, from one block long (labelled shortest copy) to eight blocks long (labelled longest copy), each ending in a coloured stop block that pairs with the template. An arrow reads the copies from short to long, and the stop letters, read in that order, spell A T G C C T A G, labelled as the new complementary strand.

**Predict.** One strand reads GATTC. What does its partner read when each partner base is written directly underneath?

> **Check your prediction.** CTAAG. G pairs with C, A with T, T with A and C with G. Biologists often write the partner strand in the opposite direction, which would reverse these letters, but here each letter sits under its match.

A **genome** is all the DNA in one set of an organism’s chromosomes, and a **chromosome** is one long DNA molecule wound with proteins. A human genome has about 3 billion base pairs.

The **Human Genome Project** was an international project, from October 1990 to April 2003, that read the order of bases in human DNA. It mostly used improved Sanger sequencing and produced a **reference sequence**, a standard human DNA sequence used for comparison.

The second film drew the reference as a ruler for lining up other sequences, but this ruler has limits. It was pieced together from several donors, with about 70 percent from one donor, and it still had gaps. Human chromosome 22, published in 1999, was the first human chromosome sequenced. Its sequence covered the gene-rich part and still had gaps.

In 2022 a team reported a human sequence with no gaps in the 22 numbered chromosomes and the X, and a finished Y followed in 2023. In 2023 another team released a draft “pangenome” built from 47 people, to show how human DNA varies. The main features of the double helix model still hold, though later measurements refined details such as how many base pairs make one turn. Sequencing added the letters, how many there are, and how they differ between people.

**Key idea:**

Sequencing reads the order of DNA bases, and since 1990 it has grown to read whole human genomes.

## The falling cost of a genome

| Date | Estimated cost per human-sized genome (US dollars) |
| --- | --- |
| September 2001 | 95,263,072 |
| October 2003 | 40,157,554 |
| October 2007 | 7,147,571 |
| January 2008 | 3,063,820 |
| January 2010 | 46,774 |
| April 2012 | 5,901 |
| July 2015 | 1,363 |
| May 2019 | 606 |
| August 2021 | 562 |
| May 2022 | 525 |

These values come from the National Human Genome Research Institute (NHGRI) “Cost per Genome” data table, May 2022 version (genome.gov). The table shows selected dates.

**Predict.** Which neighboring dates show the largest fall compared with the starting value of that interval?

> **Check your prediction.** January 2008 to January 2010. Of all the neighboring pairs in this table, this one fell by the largest factor.

1. About how many times cheaper was May 2022 than September 2001?
2. Compare 2019 to 2022 with 2008 to 2010. What do you notice about the recent change?

> **Check your answers.** 1. 95,263,072 divided by 525 is about 181,500, so roughly 180,000 times cheaper.
> 2. The recent change is much smaller: 606 to 525 dollars, against a fall of about 65 times from 2008 to 2010. In NHGRI’s full table, values after about 2015 go up and down from one date to the next.

NHGRI’s notes add details to the second film’s limits. These are production costs at NHGRI-funded sequencing centers, scaled to 3,000 million bases. They leave out interpreting the DNA and are not prices a person pays. Up to October 2007 they cover Sanger sequencing, read enough times to build a genome without a reference. From January 2008 they cover newer machines that read short pieces and line them up against the human reference. NHGRI notes that a genome without a reference would need more reading, so the early and late estimates price somewhat different jobs. The whole U.S. part of the Human Genome Project cost about 2.7 billion dollars in 1991 dollars, but that paid for much more than sequencing.

Claim, evidence, reasoning

Make a claim that answers the essential question: how did reading whole genomes change what biologists can study? Use this data.

Claim Your answer to the question, in one sentence.

Start with: The data show that the cost of sequencing a genome blank, so biologists can now blank.

Evidence The data that support your claim.

Start with: In September 2001 one genome cost about blank, but by May 2022 it cost about blank.

Reasoning Why the evidence supports the claim, using the science.

Start with: If each genome costs less, the same budget pays for blank genomes, which lets scientists compare blank.

> **Compare with a model answer.** The cost of sequencing a human-sized genome fell so far that biologists can now compare many genomes instead of one reference. In September 2001 NHGRI estimated about 95 million dollars per genome. By May 2022 the estimate was about 525 dollars, roughly 180,000 times cheaper. In this table, the steepest fall came between early 2008 and early 2010, from about 3 million dollars to about 47,000 dollars. With a fixed budget, a lower cost per genome pays for many more genomes, so scientists can compare DNA across many people and species. One limit: these are production costs from funded sequencing centers only, and they leave out the cost of analyzing the data. Another limit: after 2008 the estimates are for re-reading a human genome against an existing reference, not for building a new one, so part of the fall comes from a change in what was priced.

Saved in this browser only.

### Design problem

Sequencing machines read short pieces of DNA, and a computer joins overlapping pieces back together. Repeated stretches make this hard. Plan an investigation with paper strips to test whether longer pieces make it easier to rebuild a sequence that contains a repeat.

**Predict.** Will 6-letter, 12-letter or 20-letter strips be easiest to rebuild correctly around an 8-letter repeat?

> **One way to do it.** Write a 60-letter made-up DNA sequence that contains the same 8-letter stretch twice. Make several copies and cut them into overlapping strips.
>
> - Independent variable: strip length, such as 6, 12 and 20 letters.
> - Dependent variable: whether a classmate who has not seen the original rebuilds it correctly, and how long it takes.
> - Constants: the same original sequence, the same number of copies, and a cut at random places each time.
> - Control: a version with no repeated stretch, cut into 6-letter strips, to show the puzzle is solvable without a repeat.
> - Trials and sample size: at least five classmates for each strip length, each working alone.
>
> If short strips often give the wrong order around the repeat while 20-letter strips do not, longer reads help. This is one reason long-read methods helped fill gaps in 2022. A limit: paper strips have no reading errors, and real sequencers make some.

## Evaluate: check your thinking

Begin with the question, then ask what your chosen model hides. Reading the human genome took about 13 years partly because the methods were slow and costly. Newer technology plus a reference cut NHGRI’s estimate from about 95 million dollars in 2001 to about 525 dollars in 2022. These estimates price somewhat different jobs, but the fall helps explain why biologists can now compare many people and species.

**Science practices you used:** comparing models and naming their limits, tracing how evidence built a model, reading a data table and calculating a ratio, and planning a fair test with repeated trials.

---

From STEMhop (https://stemhop.com), free lessons by Christopher Deppmann. Lesson text © the authors. Practice questions: https://stemhop.com/lessons/discovering-dna-structure/
