Why Insertion Are Hard To Find By The Alignments Of Reads Than Deletion ?

Question

[jack]

Hi,

Does anybody know why insertions are hard to find by the alignments of reads than deletion?

[Pavel Senin]

what do you mean exactly here? I think that both phenomena are similar in terms of "finding" through the alignment procedure.

Answer 1

[zam.iqbal.genome]

Insertions by definition contain sequence that is not there in the reference. So one argument is simply that if a read has x% of inserted sequence, it only has 100-x% of sequence that can match the reference, making it harder to map. Long insertions, longer than the read, result in reads which do not map anywhere.

[Pavel Senin]

agree, that is a valid answer; long insertions, that can't be covered by a read, would be impossible to discover through the alignment procedure.

Answer 2

[Chris Whelan]

Insertions are also more difficult to find because you can't use the paired-end mapping technique if the insertion is too long. If two reads in a pair come from either side of a deletion, you can identify the deletion by looking for a longer distance between mappings of paired reads than you'd expect, and this technique is not limited in the size of deletions you can find. Insertions, on the other hand, will cause paired reads to map closer to one another than you would expect, and so if the size of the insertion is larger than the distance between paired reads in your fragment library, one of the reads will end up in the insertion and be difficult to map. Therefore, you can't use the paired-end mapping technique to find insertions larger than the fragment size of your library.