BBmerge histogram of insert size

Question

[Sharon]

Hi All

I used BBmerge to get a histogram of insert sizes in my sample. I have paired reads (2 X150) in the fastq files. I found that 45% of my reads in that sample have insert size less than 150. Average insert size is 158. Is it okay that 45 % of the reads have lower insert size than a single read.

Is a problem with library preparation ? Is there any biological or technological reason to prevent having longer insert size for Human. Like can we argue with company to get better longer insert sizes in the next samples?

Thanks

Answer 1

[GenoMax]

Did you make the libraries or did the sequence provider make them? If you did the libraries then you have no option but to remake them if you are unhappy with the insert sizes. If sequence provider made the libraries then you should look through their FAQ/agreement you had with them to see what their guarantees are for quality of libraries/data. Depending on that you may be able to make a case for getting the libraries remade/re-sequenced. Especially if the data is not serving its purpose of sampling larger genomic areas.

Ultimately this all would be dependent on initial quality of materials you had generated. If that material was crappy then you can't expect the libraries to come out any better.

[Sharon]

Thanks genomax so much. I will check with the biologists in our lab who make the library and the contract of this. but I am just confirming whether 45% low insert size than a single read length is not good? It makes adapters go through the reads, and makes long overlap which makes paired reads less informative or as informative as single reads, right?

[GenoMax]

While the libraries are probably good (I assume they pass QC/align well) they are not serving their purpose (if that was to sample a larger area of the genome). So the answer is yes to your last question. Making good libraries is a bit of an art and does require experience that one gains over time.

[Sharon]

Thanks so much genomax :) Much appreciated !