Is it possible to retrieve the original reads from a BAM file?

Question

[elcortegano]

For testing purposes, I'm interested into work with a small limited number of reads that align well to a given genomic region, also small, so computation time and memory requirements are as low as possible.

This could be done (I guess) if from the alignment BAM file one could associate the aligned reads with specific reads in the original FASTQ file.

I don't know if this is even possible. So far, I've found nothing.

[ATpoint]

Yes, reads can be extracted using e.g. samtools fastq. It is recommended to randomize the BAM before doing that as many alignment tools expect random fastq order (for paired-end data). Could be don with samtools collate or samtools sort -n followed by samtools fastq. For other threads on this please use the search function, there are many.

[cschu181]

Two questions:

1. What will samtools fastq do with hard-clipped reads/secondary alignments? Output shortened reads? I assume some kind of filtering should be applied before.

2. Do you have any reference for the random fastq order requirement? I have not heard of this before. That would be good to know for certain applications...

Edit: formatting

[GenoMax]

Yes. You can filter your BAM file to separate aligned reads and then convert them back to fastq.

You can first filter your BAM with samtools view region to get the region you need.

Then use reformat.sh from BBMap suite to retrieve reads:

reformat.sh in=your.bam out1=R1.fq.gz out2=R2.fq.gz mappedonly=t pairedonly=t primaryonly=t

If you have single end data then just use out=read.fq.gz instead.

[WouterDeCoster]

So you would like to convert bam to fastq? Have you googled for that?

[elcortegano]

Not with that words, my mistake (I'm not native speaker). Now I 've found a way with bedtools, thank you

[ATpoint]

I'm not native speaker

Don't worry. Most here aren't. Neither WouterDeCoster nor me are.

[bazghazia7]

where you able to retrieve your original reads file frombam files? Can you share the processs and commands? I really need help on it I mixed up my data and need to retrieve from bam files the raw reads.

[ATpoint]

There are multiple suggestions on how to do this in this thread. Suggest you pick one and try it.

Answer 1

[elcortegano]

Ok, I got a way to do it. You can find it in the following link: https://seqome.com/convert-bam-file-fastq/.

There it describes how bedtools bamtofastq can deal with the task. It will only require a sorted BAM file as input, eg:

bedtools bamtofastq -i input.bam -fq output.fq -fq2

[ATpoint]

name-sorted BAM if it is paired-end.

Answer 2

[ctseto]

You can pipe them out using samtools and the sam flags 64 and 128 for read 1 and read 2, and tune the extractables more carefully with convoluted combinations of the SAM flags.