Mapping variants to a reference genome discosnprad

Question

[jomare1188]

Hi guys

i want to map the variants produced by DiscosnpRad to a reference genome i have , i know DiscoSnp++ can do this with the -G flag , but it seems that DiscosnpRad can't. is there someway to map the DiscosnpRad variants to a reference genome?

thank you

Answer 1

[pierre.peterlongo]

Hi there,

Thanks for your question et remarks.

About the -G flag. Indeed, we removed the variant mapping of ref process in the discoRad tool. This is a matter of time before we propose this back in discoRad. By the way, you may still use the run_VCF_creator.sh script on your .fa file generated by discoRad. This while generate the desired .vcf

./scripts/run_VCF_creator.sh -G reference_genome.fa -p discoRad_k_31_c_auto_D_3_P_5_m_0_coherent.fa -o my_result.vcf

About the comparison reference-free / mapping approaches Currently, in my opinion, the pro/cons of each approach is not that straightforward and highly depend on the reference genome quality and its genomic distance with the species studied (not talking about computation time and resources). Moreover, in our experiments, even on simulated data, both reference-free and mapping-based approaches predicted specific true positive variants.

Pierre

Answer 2

[h.mon]

If DiscoSnpRad outputs the SNPs and its flanking sequences (say, a 50-plus base-pairs fragment), you may map these reads with Bowtie / Bowtie2 / BWA.

However, from my reading of the literature, in general approaches using a reference genome are a lot more sensitive than reference-free approaches, calling at least 1.5x more SNPs. I don't know if this is true for DiscoSnpRad in particular, but I believe it probably applies to it as well. As you have a reference genome, why don't you use for SNP calling?

[jomare1188]

i'm doing the same analysis with stacks and reference genome (in the traditional way), but since the computational requirements and runing times in discosnp are so low i wanted to do the same (in my laptop) and compare :D