Samtools: Question about filtering BAM file using flag

Question

[SDin]

Hi there, I am trying to filter BAM file with their 'flag' column. I am a little confuse about the meanings of, for example,

samtools view -f 4 -F 264 .....

I check the documentation, and notice

-f means 'what I want' and   '4'    means 'read unmapped'
-F means  'wipe off'   and  '264'   means 'mate unmapped + not primary alignment'

My question is:

If a sequence's flag is 12, will it be extracted by '-f 4'? If a sequence's flag is 8, will it be wiped off by '-F 264'?

I am confuse about the mechanism of this code, which is unfortunately not clear in the documentation.

Many thanks.

[GenoMax]

See SAM Format site for explanation of the flags.

[ATpoint]

Also see Bitwise Flag Explained

Answer 1

[i.sudbery]

The flags are numbers in base-2. Thus a better way to think about the flags is their binary encoding.

The binary encodings of the values you mentioned are:

  4: 0 0 0 0 0 0 1 0 0
  8: 0 0 0 0 0 1 0 0 0
 12: 0 0 0 0 0 1 1 0 0 
264: 1 0 0 0 0 1 0 0 0

Here we can easily see that 12 is composed of 8 (mate unmapped) and 4 (read unmapped), while 264 is composed of 256 (not primary alignment) and 8 (mate unmapped).

As -f means retain only reads with all specified flags set, a read with the flag 12 will be retained by -f 4 because a read with flag 12 has its 4 flag set. As -F means retain only reads with none of the specified flags set, a read with flag 8 will be removed by -F 264 because 8 is one of the flags specified by 264.

[i.sudbery]

of course -f 4 -F 264 will exclude a read with flag 12 because means that the 8 flag is set, and the eight flag is one of the flags that -F 264 instructs samtools to exclude.

[SDin]

I got it. Many thanks for your kind help!

Answer 2

[Friederike]

The flag values are always precise and unique because they are numerical representation of the _sum_ of the (numerical) answers to the 12 questions such as "Is the respective read mapped? Paired? ..."

See i.sudbery's answer for the technical details and perhaps play around with this page to get a better feeling for what's going on.