Planning a PacBio or Nanopore project — or troubleshooting a hybrid assembly? Our senior bioinformaticians help with genome assembly tuning, Iso-Seq isoform validation, methylation detection, and cross-platform integration — all tailored to your specific data and goals. Request a free consultation →
In recent years, long-read sequencing technologies such as PacBio SMRT, Oxford Nanopore, and the newer PromethION or Sequel II systems have become increasingly used by many research groups. Compared to short-read sequencing, long reads can provide clear advantage for solving repeat regions, large structural variants, full-length RNA isoforms, and also native base modifications. However, from my experience, proper data analysis is not always straightforward, and many pitfalls may occur if one applies standard pipelines directly without careful adjustment.
Below I share some personal observations and practical suggestions, which may be useful for researchers who plan to perform long-read genome assembly, Iso-Seq data analysis, methylation detection, or hybrid genome assembly support.
In my view, long-read genome assembly has significantly improved over the past ten years, but it is still not automatic. Even for high-quality PacBio HiFi genome assembly, parameters must be adjusted according to the specific organism, genome complexity, and expected coverage. For example, the same pipeline settings for bacterial genome may not be suitable for plant genome, which often has high repeat content and polyploidy.
Commonly used tools include Flye, Canu, Miniasm, and HiCanu for draft assembly, and Racon, Medaka, Pilon for polishing. Many people think that more polishing is always better, but over-polishing may remove real variants or merge haplotypes incorrectly.
For structural variant detection, it is usually recommended to use minimap2 for mapping, together with Sniffles, Longshot, or SVIM for SV calling. One must check the minimum read support, alignment identity, and filters carefully; otherwise, false positive or negative SV calls can happen.
We have assisted several projects where, after hybrid assembly, unexpected duplications or missing chromosome ends were found, which were caused by unbalanced polishing or misaligned short reads.
When people want to see the real full-length structure of RNA transcripts, PacBio Iso-Seq data analysis and Nanopore direct RNA data interpretation are very helpful. Compared to short-read RNA-seq, which can only infer exon junctions indirectly, long reads directly capture the whole transcript from 5' to 3' end, including alternative start sites and polyadenylation sites.
However, many users believe that once the sequencing is done, they can directly obtain high-quality isoforms without much post-processing. In reality, steps like clustering, collapsing, chimera removal, and novel isoform validation are still necessary. Pipelines such as FLAIR, TALON, FLAMES, and IsoQuant can help in different situations.
One common point is that highly expressed genes may dominate, so low-expression isoforms may have insufficient read support. Therefore, it is good practice to compare discovered isoforms with reference annotations like GENCODE or RefSeq, and confirm important novel structures by RT-PCR or Sanger sequencing when possible.
A big advantage of long reads is that they can detect methylation and some other base modifications natively, without extra chemical treatment. For PacBio base modification detection, the software uses polymerase kinetics signal, while for Nanopore methylation calling analysis, signal-level models like Megalodon, Remora, or Nanopolish are common choices.
In my opinion, these tools provide good starting point, but the modification calling is not 100% precise, and the detection level often depends on the quality of raw reads, signal noise, and basecaller version. It is recommended to calibrate thresholds carefully and, if the biological conclusion is important, to validate with another method, such as bisulfite sequencing or methylation-specific PCR.
For RNA modifications, some teams also try to detect m6A or A-to-I editing using Nanopore direct RNA, but this area is still under active development and may not yet give very reliable quantitative results.
Hybrid genome assembly combines short reads and long reads, and can achieve better balance between contiguity and base accuracy. Many bacterial genome projects use Nanopore Illumina hybrid genome assembly, and for larger genomes, PacBio Illumina hybrid assembly support is very common.
However, from my experience, many mistakes happen here: for example, using too few short reads for polishing a large genome, or applying polishing too many times which removes real sequence diversity. Also, when merging data from different runs or instruments, one must be careful about adapter trimming and barcode demultiplexing, as leftover adapters can cause contig misjoins.
Tools like Unicycler, MaSuRCA, or hybrid mode of SPAdes are popular, but they all need some parameter tuning based on read length, coverage ratio, and expected genome size.
In summary, long-read genome assembly support, Iso-Seq data analysis, long-read methylation detection, and hybrid genome integration can provide deeper insights compared to standard short-read analysis, but they require careful planning and flexible pipeline design. From my personal experience, spending extra time on parameter tuning and QC always saves more time and confusion later.
If you have PacBio or Nanopore data, or plan to generate such data, and need an experienced team to help with analysis, you are welcome to discuss with us. We are happy to share our experience and provide suggestions suitable for your specific research goal.
About the author: Zack Tu holds a B.S. in Biochemistry, an M.S. in Software Systems, and a Ph.D. in Pharmacology. With over two decades of experience in computational genomics, Zack has supported a wide range of academic and industry research projects involving high-throughput sequencing and complex data integration. He previously led the core bioinformatics infrastructure at the University of Minnesota’s research sequencing center and contributed to developing clinical genomics workflows for rare disease diagnostics. Since joining AccuraScience in 2013, he has worked extensively with next-generation and third-generation sequencing platforms, providing guidance on genome assembly, variant detection, transcriptomics, epigenetics, and multi-omics data interpretation. His work emphasizes practical, reproducible pipelines and thoughtful analysis tailored to each research question. More at Our Team page.Need help with long-read or hybrid data? Learn how we can help, or check out our FAQ.
Send us an inquiry, chat online (during business hours 9–5 Mon–Fri U.S. Central Time), or contact us another way!