Difference between revisions of "BamUtil: convert"

From Genome Analysis Wiki
Jump to: navigation, search
(add recovery section)
(BAM File Recovery: better grammar and more text)
Line 64: Line 64:
 
= BAM File Recovery =
 
= BAM File Recovery =
  
A BAM file that has been corrupted, or truncated due to a copy or disk problem can often be partially recovered.
+
A BAM file that has been corrupted or truncated due to a copy or disk problem can often be partially recovered.
  
 
Both the BGZF format and binary BAM format have enough information to scan forward and resynchronize the input data.  While some data will be lost, substantial recovery can often be done.
 
Both the BGZF format and binary BAM format have enough information to scan forward and resynchronize the input data.  While some data will be lost, substantial recovery can often be done.
Line 80: Line 80:
  
 
Currently, no statistics are printed as far as how many BAM records are recovered, but subsequent tests can readily be done on the resulting file to determine the quality of recovery.
 
Currently, no statistics are printed as far as how many BAM records are recovered, but subsequent tests can readily be done on the resulting file to determine the quality of recovery.
 +
 +
In real cases, we have recovered better than 94% of reads from a set of severely damaged files (numerous 64K chunks of a RAID were lost), and better than 99.9% recovery from a moderately damaged file (3 disk pages were corrupt).
  
 
= Usage =
 
= Usage =

Revision as of 13:47, 4 October 2011


Overview of the convert function of bamUtil

The convert option on the bamUtil executable reads a SAM/BAM file and writes it as a SAM/BAM file.

The executable converts the input file into the format of the output file.

It has options to allow for the conversion of the sequence to/from '=' from/to the actual bases by using the reference sequence.

If you want to convert a BAM file to a SAM file, just call:

<pathToExe>/bam --in <bamFile>.bam --out <newSamFile>.sam

Don't forget to put in the paths to the executable and your test files.

Parameters

    Required Parameters:
        --in        : the SAM/BAM file to be read
        --out       : the SAM/BAM file to be written
    Optional Parameters:
	--refFile   : reference file name
        --noeof     : do not expect an EOF block on a bam file.
        --params    : print the parameter settings
        --recover   : attempt to recover the input bam file.
    Optional Sequence Parameters (only specify one):
	--seqOrig   : Leave the sequence as is (default & used if reference is not specified).
	--seqBases  : Convert any '=' in the sequence to the appropriate base using the reference (requires --ref).
	--seqEquals : Convert any bases that match the reference to '=' (requires --ref).

Sequence Representation Parameters

The sequence parameters options specify how to represent the sequence if the reference is specified (refFile option). If the reference is not specified or seqOrig is specified, no modifications are made to the sequence. If the reference and seqBases is specified, any matches between the sequence and the reference are represented in the sequence as the appropriate base. If the reference and seqEquals is specified, any matches between the sequence and the reference are represented in the sequence as '='.

Examples

ExtendedCigar: SSMMMDDMMMIMNNNMPMSSS
Sequence:      AATAA  CTAGA   T AGGG
Reference:       TAACCCTA ACCCT A
Sequence with Orig:   AATAACTAGATAGGG
Sequence with Bases:  AATAACTAGATAGGG
Sequence with Equals: AA======G===GGG
ExtendedCigar: SSMMMDDMMMIMNNNMPMSSS
Sequence:      AATGA  CTGGA   T AGGG
Reference:       TAACCCTA ACCCT A
Sequence with Orig:   AATGACTGGATAGGG
Sequence with Bases:  AATGACTGGATAGGG
Sequence with Equals: AA=G===GG===GGG
ExtendedCigar: SSMMMDDMMMIMNNNMPMSSS
Sequence:      AAT=A  CT=GA   T AGGG
Reference:       TAACCCTA ACCCT A
Sequence with Orig:   AAT=ACT=GATAGGG
Sequence with Bases:  AATGACTGGATAGGG
Sequence with Equals: AA======G===GGG
ExtendedCigar: SSMMMDDMMMIMNNNMPMSSS
Sequence:      AA===  ===G=   = =GGG
Reference:       TAACCCTA ACCCT A
Sequence with Orig:   AA======G===GGG
Sequence with Bases:  AATAACTAGATAGGG
Sequence with Equals: AA======G===GGG

BAM File Recovery

A BAM file that has been corrupted or truncated due to a copy or disk problem can often be partially recovered.

Both the BGZF format and binary BAM format have enough information to scan forward and resynchronize the input data. While some data will be lost, substantial recovery can often be done.

When a file has bad blocks in it, normal copy commands (cp) will truncate the file at the point of disk read failure. To recover the maximum amount of data possible, use the dd command with the conv=noerror option.

So a normal use case for recovery would look this this:

# dd if=brokenbamfile.bam of=/tmp/brokenbamfile1.bam conv=noerror bs=4k
# bam convert --recover --in /tmp/brokenbamfile1.bam --out /tmp/brokenbamfilerecovered.bam

Note, you will of course need to output the result file to a known good filesystem.

Currently, no statistics are printed as far as how many BAM records are recovered, but subsequent tests can readily be done on the resulting file to determine the quality of recovery.

In real cases, we have recovered better than 94% of reads from a set of severely damaged files (numerous 64K chunks of a RAID were lost), and better than 99.9% recovery from a moderately damaged file (3 disk pages were corrupt).

Usage

./bam convert --in <inputFile> --out <outputFile.sam/bam/ubam (ubam is uncompressed bam)> [--refFile <reference filename>] [--seqBases|--seqEquals|--seqOrig] [--noeof] [--params]


Return Value

Returns the SamStatus for the reads/writes.

Example Output

Number of records read = 10
Number of records written = 10