Biology · Genomics, Omics & Bioinformatics
Sequencing, reads, assemblies and data quality
Four short records contain sixteen letters, but the imagined reference segment contains only eight positions. Some positions were read more than once. Counting all the letters therefore does not tell us the length of the segment, how many organisms were sampled or whether the reconstruction is correct.
Gene Expression: RNA & Proteins explains how sequence, RNA and protein are related. Phylogeny, Classification & the Tree of Life introduces ancestry and the interpretation of trees. Research Design, Data & Scientific Practice explains sampling, independent units and the limits of comparisons. Their English chapters are available in this working manuscript; publication remains pending. The examples here use letter strings, inclusive coordinates, averages, fractions and ordinary graph reading. Each convention is introduced before it is used.
Back to the Biology mapBy the end, you can
Distinguish a read record, an assembled sequence and a biological genome, and interpret coverage without treating it as a guarantee of correctness.
A sequence record represents an observation
A sequencing read is a record of base calls from an observed fragment. Its letters describe an inferred order, usually with additional information about measurement quality. A read is not the whole genome. Multiple reads can come from different regions, overlap the same region or contain errors. Their biological origin and measurement history belong to their interpretation.
An assembly is a reconstruction that places sequence information into longer representations. A contig represents a contiguous assembled sequence; other representations can link pieces while retaining gaps or uncertainty. The biological genome is the genetic material being represented, rather than the file or reconstruction itself. An assembly may omit regions, combine alternatives or contain an error even when its file is perfectly readable.
In the original eight-position example, a fixed reference is supplied so that we can count where each read lies. All four records have length four. Their overlap explains why sixteen recorded bases cover only eight reference positions. These are invented letters in one orientation, not fragments of a real organism or instructions for obtaining sequence data.
Overlapping records count the same positions
Read the diagram in words
Reads one and four cover one to four, read two covers three to six, and read three covers five to eight. Depths are 2, 2, 3, 3, 2, 2, 1, 1: total sixteen, mean two, and all eight positions covered.
| Supplied reference segment | DNA letters in the assigned orientation | Last reference position |
|---|---|---|
| Supplied reference segment | ATGCCATA | 8 |
| Assigned read record | DNA letters in the assigned orientation | First reference position | Last reference position |
|---|---|---|---|
| Read one | ATGC | 1 | 4 |
| Read two | GCCA | 3 | 6 |
| Read three | CATA | 5 | 8 |
| Read four | ATGC | 1 | 4 |
| One-based reference position | Displayed overlapping read count |
|---|---|
| 1 | 2 |
| 2 | 2 |
| 3 | 3 |
| 4 | 3 |
| 5 | 2 |
| 6 | 2 |
| 7 | 1 |
| 8 | 1 |
| Sum of all recorded read bases | 16 |
|---|---|
| Mean depth over eight positions | 16/8 = 2 |
| Fraction of supplied positions with a read | 8/8 = 1 |
Coverage is not a correctness guarantee
Original eight-position DNA-letter representation with four assigned error-free, fixed-orientation reads and known coordinates. No real genome, measurement, independent biological sample count or biological procedure is supplied.
Read depth and correctness answer different questions
Coverage must name what is being counted. Here, depth at a position is the number of displayed read records overlapping that position. Mean depth is the total of those positional counts divided by the reference length. Breadth is the fraction of reference positions covered by at least one displayed read. A mean can hide uneven coverage, so the individual positions remain visible.
The example has mean depth two and breadth one: every supplied position is covered. Yet the last two positions have one record each while the middle positions have more. A repeated record is an additional observation in this account, not proof of an additional independent biological sample. More records can reduce some uncertainty while leaving systematic errors or ambiguous placement unresolved.
Real data quality also concerns which bases were called confidently, whether records belong to the intended material and whether a reconstruction is compatible with the observations. No single average settles all these questions. Alignment to a supplied reference provides a coordinate system; it does not establish that this reference contains every sequence found in every individual.
Compatible fragments may leave more than one reconstruction
Consider only the two distinct strings ATAT and TATA. Both occur within ATATA, and both also occur within ATATATA. Their presence alone therefore leaves the repeat length unresolved. The second model records every occurrence in the two candidate strings so that compatibility is visible, rather than treating one plausible arrangement as the unique answer.
This example gives only the distinct observed patterns. It supplies no exact sampling multiplicities, end positions or independent evidence about total length. The candidates are not equally probable conclusions simply because both are compatible. They show that extra information is needed to distinguish alternatives under the stated limited evidence.
References help comparisons by providing a shared representation. A reference can be useful without being identical to every individual's genome. Collections that represent multiple sequence alternatives can expose variation missed by a single path. Their usefulness still depends on which genomes and evidence they represent; the word reference does not certify completeness or universal representativeness.
The same patterns fit different lengths
Read the diagram in words
ATAT and TATA occur at starts one and two in ATATA. In ATATATA, ATAT starts at one and three and TATA at two and four. Both candidates fit the limited presence evidence.
| Distinct observed pattern |
|---|
| ATAT |
| TATA |
| Candidate string | Candidate length |
|---|---|
| Five-position candidate: ATATA | 5 |
| Seven-position candidate: ATATATA | 7 |
| Candidate string | Distinct observed pattern | Pattern start | Pattern end |
|---|---|---|---|
| Five-position candidate | ATAT | 1 | 4 |
| Five-position candidate | TATA | 2 | 5 |
| Seven-position candidate | ATAT | 1 | 4 |
| Seven-position candidate | ATAT | 3 | 6 |
| Seven-position candidate | TATA | 2 | 5 |
| Seven-position candidate | TATA | 4 | 7 |
Compatibility does not supply uniqueness or probability
Original distinct-pattern compatibility problem in a fixed orientation. Only presence of the two distinct patterns is supplied, not read multiplicities, fragment boundaries, exact total length, probabilities or optimal assembly.
Worked example
Count positions rather than concatenated letters
The first and fourth reads cover positions one to four, the second covers three to six, and the third covers five to eight. Counting overlaps at each position gives two, two, three, three, two, two, one and one. These counts sum to sixteen.
Divide sixteen by the eight supplied reference positions to obtain mean depth two. All eight positions have a positive count, so breadth is eight divided by eight, or one. This describes the assigned record; it supplies neither an error probability nor a count of independent organisms.
For the separate repeat case, ATATA contains ATAT starting at one and TATA starting at two. ATATATA contains ATAT starting at one and three, and TATA starting at two and four. Both satisfy the distinct-pattern evidence, so selecting one length requires an additional justified constraint.
Check your understanding
Do sixteen bases across the four read records imply a sixteen-base genome segment?
Show reasoning
No. Overlapping records count some positions repeatedly. In the supplied coordinate model, sixteen read bases cover eight reference positions. A genome-length claim would require more than the sum of read lengths.
Does complete breadth across the toy reference establish a correct or complete biological genome assembly?
Show reasoning
No. It establishes that every supplied reference position has a displayed read. Errors, omitted alternatives and sequences outside that reference remain separate questions.
Why do ATAT and TATA fail to distinguish the two repeat candidates?
Show reasoning
Each pattern occurs in both candidates. The example supplies distinct patterns rather than exact origin, multiplicity or length information, so compatibility does not identify a unique reconstruction.
What to carry forward
Identify the represented object and its evidence: a read, an assembly and a genome are related but distinct.
References
Resources supporting this lesson’s explanations and examples.