2 questions on Bowtie

[eltonsky]

Hellow fellas

I rencently started working on using crossbow to process mouse genome sequence. I am from computer science background and I know quite little about biology

After some experiments and reading through the manual of bowtie, I still didn't get:
1. The command for bowtie is like:
"./bowtie index/index -p 2 --partition $partition_size -X $insert_size -m 1 -v 2 --best --strata --mm --12 --startverbose --mmsweep\"
It is using -m 1 in conjunction with "-best --strata". My understanding is when there's only 1 reportable alignment in the best alignment "stratum", this alignment will be reported.
Bowtie discards alignments when accidentally 2 or more meet the same "best" criteria. is this right?

2. In the output of bowtie, the last second column is called "hit". From the manual:
"this column contains the number of other instances where
the same sequence aligned against the same reference characters as
were aligned against in the reported alignment. This is *not* the
number of other places the read aligns with the same number of
mismatches."
I could not understand this statement. Can any biologist here give me some example of what this means?

Really appreciate if you can help me out here.
Cheers

[akundaje]

Regarding 1.

Yes thats right -m1 --best --strata will not report any reads that map (within the best stratum of alignment) to more than one location.

I am not sure where you got the --partition, --startverbose and --mmsweep parameters from? Maybe you are using some older version of bowtie? The new version certainly doesn't have them.

Regarding 2., this is how I would interpret it. Others can correct me if I am wrong.

Lets say read AGC was reported as mapping to a location whose sequence is ACC.

The parameters allowed 1 mismatch.

The "hit" column will then report the number of other locations with the exact sequence ACC to which the read (AGC) can map to but wasn't reported.

However, this is different from the number of all other locations to which the read (AGC) could map allowing the 1 mismatch. AGC could also map to locations whose sequence is AGG. However this number is not included in the "hit" column since the reported alignment is to ACC.

[eltonsky]

Thank you very much for reply akundaje!

I am not sure where you got the --partition, --startverbose and --mmsweep parameters from? Maybe you are using some older version of bowtie? The new version certainly doesn't have them.

I was working on crossbow (version 0.1.3). And the " --partition, --startverbose and --mmsweep " options are from the script "crossbow.pl" which drives crossbow running on your local hadoop cluster (in /local folder).

Regarding 2., this is how I would interpret it. Others can correct me if I am wrong.

Lets say read AGC was reported as mapping to a location whose sequence is ACC.

The parameters allowed 1 mismatch.

The "hit" column will then report the number of other locations with the exact sequence ACC to which the read (AGC) can map to but wasn't reported.

However, this is different from the number of all other locations to which the read (AGC) could map allowing the 1 mismatch. AGC could also map to locations whose sequence is AGG. However this number is not included in the "hit" column since the reported alignment is to ACC.

So if I have some exact reference sequence "AGC", say 5, then the "hit" will give me 4 in all these 5 alignment records, if they are reportable. And if there's another sequence "GGC", say also 5 of them, "hit" also give's 4 in all those 5 aligment. Am I right?

Elton

[akundaje]

I did not exactly understand your example. Just to reiterate

Your read sequence is AGC.

The line in the bowtie output for this read says that it maps a location (say chr1,20) at which the sequence is ACC.

Now the same sequence ACC (note: this is the sequence TO WHICH the read was mapped to and may not be the same as the read sequence) exists in say 10 other locations in the genome (i.e. a total of 11 locations that have sequence ACC).

Then for this line of bowtie output, the "hit" column should have a value of 10.