Skip to content
Tech Interview Prep
Technical interview guide

Genome Assembly & Sequencing Technologies

How raw sequencing reads become a genome: short-read vs. long-read platforms, de Bruijn graph assembly, and what coverage actually buys you.

Read
49 min
Practice MCQs
25
Interview Q&A
25
Edition
v3
Editorial status
Reviewed

Scope: Foundational Lander-Waterman and de Bruijn graph models; SPAdes, Canu, Flye, hifiasm, QUAST and BUSCO source publications; current Illumina, PacBio, Oxford Nanopore and NCBI guidance reviewed 2026-09-04.

Curated: · Written: · Reviewed:

25 questions

MCQ-10A 2×150 bp paired-end library is being used to scaffold an assembly that stalls at repeat boundaries. The same reads are also available as two independent single-end files. What does the paired-end relationship contribute that the single-end reads cannot?
MCQ-4A diploid plant genome is estimated at 500 Mb. A sequencing run yields 90 Gb of raw reads; adapter trimming and quality filtering remove 10% of the bases, and 5% of the remaining bases are mitochondrial and chloroplast reads that the assembly pipeline will discard. The team wants a nominal haploid depth of at least 30x from nuclear reads. What is the nominal haploid coverage the run actually delivers, and does it meet the target?
MCQ-17A short-read assembly of a 40 Mb bacterial genome is submitted as eight contigs with lengths 10, 8, 7, 5, 4, 3, 2 and 1 Mb. The pipeline's summary line reports contig N50 = 10 Mb. What is the assembly's actual contig N50?
MCQ-18A draft assembly of a 120 Mb genome (size estimated from a k-mer spectrum) contains four contigs of 55, 20, 15 and 10 Mb, totalling 100 Mb. What are the assembly's N50 and NG50?
MCQ-7A de Bruijn assembler is run twice on the same 150 bp reads from a genome that contains a 40 bp sequence repeated twice in different flanking contexts. With k = 55, the assembler produces a single contig through the repeat; with k = 21, every contig ends at the repeat's boundaries. What explains the difference?
MCQ-2A team assembles a bacterial genome from 2×150 bp paired-end reads with a 300 bp insert library. The assembly collapses a 5 kb rRNA operon that occurs in seven near-identical copies. Which property of the reads explains this?
MCQ-6A 4.0 Mb bacterial genome carries a 15 kb sequence in three copies that differ by <0.1%, separated by unique sequence. PacBio HiFi reads average 10 kb. An OLC assembler reports a single 3.97 Mb contig. What explains the missing 30 kb?
MCQ-9A 4.8 Mb bacterial genome is assembled from 150 bp Illumina reads at 60x. The assembly collapses a 5.2 kb repeat that occurs in three copies, each flanked by different unique genes, into a single copy. A colleague proposes fixing this by increasing coverage to 200x with the same library. What is the most likely outcome?
MCQ-8A 3.2 Mbp bacterial genome is sequenced with 2×100 bp reads at 20×. A 30 bp repeat occurs 5 times. k=21 yields 80 contigs, N50 45 kb, with the repeat collapsed. k=71 yields 350 contigs, N50 8 kb, with the repeat resolved. Which action is best supported by these results?
MCQ-1A team assembles a 3.2 Gb diploid plant genome that is 71% repetitive and carries about 380 near-identical 8 kb LTR retrotransposon copies. From 60x Illumina 2x150 bp reads they get 118,000 contigs with a contig N50 of 1.8 kb, the LTR family shows up as a single collapsed copy, and contigs end at its boundaries. They need contig N50 of at least 1 Mb and have budget for one more sequencing run. Which single addition is most likely to reach that target?
MCQ-3You assemble a 4.8 Mb bacterial genome. Unpaired 2×150 bp short reads at 100× yield 37 contigs; 12 breaks sit inside a 6.2 kb IS element present in 8 copies, each flanked by different unique sequence. You add Nanopore reads with N50 18 kb at 30× coverage. Which outcome is expected after hybrid assembly?
MCQ-11A diploid plant is sequenced to 40×. At a heterozygous SNP, the reference allele has 20× coverage and the alternate allele has 7× because of local GC bias. The assembler's graph-cleaning pass removes branches with coverage below 10×. What is the most likely outcome at this locus?
MCQ-15You assemble a diploid sample at 30x coverage with a de Bruijn graph (k = 55) from short reads. At one locus the graph shows a bubble between two unique flanks: branch A is 1,150 bp at 15x mean k-mer coverage, branch B is 1,140 bp at 16x, and branch C is 180 bp at 2x. The flanking unique sequence sits at 30x. Which branches should the final assembly keep?
MCQ-5You assemble a metagenome from two independent 60× Illumina libraries with k=31 (per-base error rate ~0.1%). The dominant genome's k-mers sit at 58–62×. Two low-abundance features appear: (A) a 31-bp dead-end tip whose k-mers are at 1–2× and appear in only one library; (B) a 600-bp unbranched path whose k-mers are at 2–3× and appear in both libraries. What is the correct call?
MCQ-14A haplotype-resolved assembly reports a 2 Mb phase block containing 100 heterozygous SNVs. The pipeline's switch error rate is 2% per heterozygous site: between each adjacent pair of sites, the phase is flipped relative to the previous site with probability 0.02, independently of the other sites. A downstream analysis uses the first and last SNV in the block as a linked marker pair. What is the probability that the assembly reports the correct phase between these two SNVs?
MCQ-12A 40 Mb haploid fungal genome is assembled from long reads, then scaffolded with Hi-C. The final report gives total length 41.2 Mb, contig N50 250 kb, scaffold N50 3.1 Mb, and 3.6 Mb of the total as N. Which conclusion does the report support?
MCQ-13A team assembles a bacterial genome from 60x Illumina 2x150 bp reads with SPAdes, then runs three rounds of Pilon polishing against the same read set. After round 1 the mismatch rate against a finished reference drops from 42 to 9 per 100 kbp; after rounds 2 and 3 it stays at 9 per 100 kbp, but the assembly's total length grows by 180 kbp and a 12 kbp prophage region that was fully spanned by reads is now split into two contigs. What does this outcome indicate about the polishing run?
MCQ-25A team assembles a bacterial genome from 60× Nanopore reads, then runs three rounds of short-read polishing with 100× Illumina reads. In homopolymer runs longer than 8 bases, the final consensus is consistently 1–2 bases shorter than the true sequence, and the same shortening appears after every round. What is the primary reason?
MCQ-24Your team has 25x PacBio HiFi reads (about Q30, so the read k-mer histogram splits into a count-1 error peak and a solid peak near 25x) and a haplotype-phased de novo assembly of the same 3.1 Gb genome. There is no reference genome, and the QC pipeline does no read-to-contig alignment. A reviewer asks for two numbers from that one k-mer table: how much of the genome the assembly recovered, and a Phred-scaled per-base accuracy of its consensus. Which computations produce those two numbers?
MCQ-21A diploid genome was assembled with a short-read assembler, and the resulting contigs are being validated by mapping the original paired-end reads back with bwa-mem. One 40 kb contig shows uniform 30x depth across its first 35 kb, then a 5 kb terminal segment where depth drops to 4x, 60% of reads have soft-clipped bases at the junction, and 12 read pairs span the junction with an insert size 3x the library mean. What does this pattern most likely indicate about that terminal segment?
MCQ-20A diploid fish genome has a flow-cytometry haploid estimate of 800 Mb. You assemble 1.24 Gb with N50 3.2 Mb and run BUSCO against actinopterygii_odb10: C:96.8% [S:22.4%, D:74.4%], F:1.6%, M:1.6%. Coverage is uniform at ~55× across the assembly, and no second species' reads are present in the library. What does the duplication most likely reflect?
MCQ-19A diploid human sample is assembled from 35x PacBio HiFi plus Hi-C into an 8.6 Mb contig. Hi-C shows the first 5.2 Mb and the last 3.4 Mb each contacting their own chromosome territory, with contacts between the two segments at background level. Coverage is flat at 31x across the whole contig. Of the HiFi reads spanning the 5.2/3.4 junction, 71% have their two halves aligning to different chromosomes at >99% identity over >10 kb. The two segments match chr3 and chr11 of GRCh38, and the sample is a normal diploid with no known translocation. What should be done with this contig?
MCQ-22Flye 2.9 assembles a 3.2 Mb bacterial genome from 62× Nanopore reads into a single contig that it flags as circular. The contig's first 1,150 bp match its last 1,150 bp exactly, and that terminal region has 124× mean depth while the rest of the contig sits at 62×. Every read that crosses the junction lies entirely within the 1,150 bp repeat. What does this evidence support?
MCQ-23A 15-species mock community is sequenced to 10 Gbp of 150 bp reads. One member has a 4 Mb genome at 0.05% relative abundance; a read-level survey confirms its single-copy marker is covered at ~1.2x. You assemble with a single-isolate assembler whose minimum k-mer coverage threshold is 5x. The assembly contains no contigs from that genome, and one 60 kb contig has its left half matching species A and its right half matching species B, which share a 3 kb conserved region covered at 40x in A and 6x in B. Which explanation accounts for both observations?
MCQ-16A Klebsiella pneumoniae isolate was grown from a single colony and sequenced to a 46× median coverage. The assembly is 5.4 Mb in 68 contigs (N50 210 kb). One 38 kb contig has GC 63.1% against an assembly median of 57.4%, and its best hit is 78% nucleotide identity over 2.1 kb to a gene in a soil Alphaproteobacterium. Its own mean coverage is 44×. It has zero reads in the DNA extraction blank and zero in the no-template control, both sequenced from the same batch. Thirty-one read pairs span from this contig into the chromosomal cps (capsule) locus. How should this contig be classified?