Explanation of BWA index files

[getVictor]

bwa index generates a bunch of files:
.rpac, .amb, .ann, .pac, .bwt, .rbwt, .rsa, .sa

Is there an explanation for what each file is for? What information is each supposed to represent?

[xied75]

Knowing each file type is interesting task, one thing for sure is you can't miss any to let BWA work.

Seems you are using an older version.

.amb is text file, to record appearance of N (or other non-ATGC) in the ref fasta.
.ann is text file, to record ref sequences, name, length, etc.
.bwt is binary, the Burrows-Wheeler transformed sequence.
.pac is binary, packaged sequence (four base pairs encode one byte).
.sa is binary, suffix array index.

[getVictor]

So .pac is a compressed version of the original reference FASTA file, correct?

[xied75]

It is, 1/4 of fasta size, very easy to decode, just take a byte, 0x 00 00 00 00 gives you AAAA.

A 0 (00) C 1 (01) G 2 (10) T 3 (11)

[getVictor]

Do AMB and ANN acronyms stand for something?

ANN sounds like for annotation.

[xied75]

Given I'm not the author of BWA, my guessing is as good as yours. Since all non ATGC will be replaced by randomly generated ATGC during the INDEX, I guess AMB stands for ambiguous.