Hi researchers,
I need your help as I am not sure if my thoughts are correct regarding the usage of a repeat-masked genome as reference for read mapping and SNP calling. I am not familiar with analyses on a whole genome level and I was wondering if you have good reasons why or why not using a (hard-)masked genome? It will be highly time consuming running it with both genomes.
I thought using a masked genome would reduce the computational power. I map short reads to detect SNPs for Population Genomcis Study (detecting population structure). I have many reads per sample and a 2.4Gbp genome. If I am interested in SNPs does it matter if I cover the repetitive regions? Does is have any effect on the mapping quality? If reads would map to masked regions, but instead map incorrectly to another region, can I filter them out by mapping quality?
Alternatively, I thought about using the non-masked genome but removing the scaffolds that are only repeats (or with other words would be 100% masked in the masked genome).
I appreciate your feedback. What are your thoughts? What would be accepted by Journals?
Have a great evening. I am looking forward to you ideas and arguments.
Julia
Hi Julia,
Thank you for reaching out with your questions. I understand your concern about the potential computational savings from using a repeat-masked genome. However, masking repetitive regions can lead to significant drawbacks for your SNP detection and population genomics study. By masking these areas, you might miss important genetic variations that are crucial for understanding population structure. Additionally, masking can cause reads from repetitive regions to either not map or map incorrectly to other parts of the genome, which can introduce biases and reduce the accuracy of your SNP calls. Even if you filter based on mapping quality, some incorrectly mapped reads might still affect your results.
Instead of using a masked genome, I recommend using the unmasked reference and handling repetitive regions through other methods. For example, you could remove scaffolds that are entirely repetitive or use advanced mapping tools and parameters optimized for dealing with complex genomic regions. This approach allows you to retain more valuable data while managing computational resources effectively.
Journals typically prefer methods that preserve as much genetic information as possible while maintaining accuracy, so demonstrating that your approach carefully handles repetitive regions without unnecessary masking will likely be well-received. Providing a clear rationale and supporting your choices with established best practices will strengthen your study for publication.
I hope this helps clarify your approach. Best of luck with your research!
Best regards,
Pan
Dear Jin-Min,
Thank you for your recommendation. :)
I guess I will continue with the removal of scaffolds that are entirely repetitive and see how it works.
In this context I might have another question. My first mapping trial received an output SAM file of 34GB. The genome is 2.4Gbp long (2.3GB file) and I mapped paired reads (8.2 and 8.3 GB file) against it. I used 64 threads and after 24h the run was not completed. This was also the initial reason why I thought about using the masked-genome.
I appreciate any recommendation or comment. I try to understand what I have to expect. Thank you.
Have a great one.
Julia
- Badges
- Science topic