That’s a great question — it shows you’re really thinking deeply about how this works, which is impressive for a third-year undergrad! Let me walk you through it in a simple, relatable way using some analogies to help it

Why 600bp Gaps in Paired-End Sequencing Are Actually Great

It might seem like you're missing data with a 600bp gap between paired-end reads, but you're not! Think of it like mapping a city with overlapping helicopter photos — you don’t need every inch, just enough overlap from different angles to piece it all together.

Why 600bp Is the Sweet Spot:

That 600bp gap is by design — it improves mapping accuracy, structural variant detection, and genome coverage without sacrificing quality or cost. Perfect for finding tricky, causative variants.

The above answer is factually correct, but I have the feeling it didn't really answer your question (and it reads like AI).

While each sample in a particular lane will have ~300 bp of sequence + 600 bp gap + ~300 bp sequence, there are lots of samples from that region & not all will start & stop in the same place.

300 bp is typically enough to start aligning those short reads with each other. Knowing that there are ~600 bp between the ends helps with the alignment. Basically, can can "well, these 300 bp are here & the other 300 bp must be 600 bp away". Narrows down the list of possible ways to line up the data.

I made this as a mock up, there are 2 bp + 4 unknown + 2 bp

You have 3 sequences

Sequence 1: AT NNNN GC

Sequence 2: TA NNNN CC

Sequence 3: AA NNNN CT

You'd get a consensus of:

ATAA N GCCT

With enough reads, you can figure out what is in the "gaps".

I hope that helps! I loved learning about this when I was a 3rd year too.

If you want better coverage you need to increase the read depth, not reduce the space between paired reads. Two reads in a pair. each with 300bp, will give you 600bp of coverage no matter how far apart they are spaced (or less if they end up overlapping, as paired reads often do - overlapping reads give a false estimate of read depth since they really shouldn't be counted separately since they are sequencing the same fragment). What paired ends help you with is phasing - if you have two variants in a sequence you want to know which ones are part of the same allele. By spacing the pairs reasonably far apart from each other, you're more likely to be able to see which variants segregate together. With single reads or highly overlapping reads, you're only able to phase variants that occur on the same read or otherwise extremely close to each other, and you end up with much lower average phased length.

Adrienne Horrell , In paired-end sequencing, each DNA fragment is sequenced from both ends, producing two reads per fragment. For example, if you have a 1200 base pair (bp) fragment and sequence 300 bp from each end, there will be an unsequenced gap of about 600 bp in the middle. This gap doesn’t mean that the sequence information in that region is lost; instead, the paired reads serve as anchors on either side of that gap, allowing bioinformatics tools to infer what’s likely in between.

Having an insert size that results in a 600 bp gap is often preferred because it improves the accuracy of read mapping to a reference genome. Short inserts can cause both reads to fall within repetitive or ambiguous regions, making it hard to place them uniquely. Longer inserts, like those with a 600 bp gap, span larger genomic regions and help ensure that reads map uniquely on both sides, which is critical for confidently detecting variants.

Additionally, paired-end reads with a reasonable gap size are invaluable for detecting structural variants such as insertions, deletions, or rearrangements. The distance and orientation between paired reads can indicate if the genome in that sample differs structurally from the reference, which might point to causative variants affecting the phenotype you’re studying.

While it might seem intuitive to want to sequence every base directly for maximum coverage, this isn’t necessary. Instead, sequencing many fragments with overlapping inserts allows computational methods to reconstruct the full sequence across the region. This approach balances cost, efficiency, and accuracy. By sequencing both ends and knowing the approximate insert size, variant callers can infer the missing middle region from the collective data.