BAM/SAM to FASTA conversion

Question

[biolab]

Hi everone,

I searched Biostars for BAM/SAM to FASTA conversion method, and found the tools EMBOSS Picard could do this (Convert Bam File To Fasta File). I am wondering is there any perl script to accomplish this work? I should note that my seq data is strand-specific seq, so the command samtools view filename.bam | awk '{OFS="\t"; print ">"$1"\n"$10}' may not be very good.

I appreciate any of your solutions. Thanks a lot!

Answer 1

[Carambakaracho]

For future record: samtools versions >1.3 can convert bam to fasta directly via samtools fasta

samtools --help

Program: samtools (Tools for alignments in the SAM format)
Version: 1.4 (using htslib 1.4)

Usage:   samtools <command> [options]

Commands:
  [...]
  -- File operations
     [...]
     fastq          converts a BAM to a FASTQ
     fasta          converts a BAM to a FASTA
  [...]

Answer 2

[Istvan Albert]

The advice in the post you link to is greatly outdated. One of the reasons we should revisit these posts. There are far easier/faster ways to accomplish the same task.

The latest versions of samtools have the bam2fq command. Chain that to seqtk seq like so:

samtools bam2fq input.bam | seqtk seq -A > output.fa

[biolab]

Hi Istvan, thank you very much for your answer. It's much helpful for my work. I just have two more minor questions: (1) my samtools version is Version: 0.1.8 (r613). I did not see samtools bam2fq function, so I need to update samtools? (2) Additionally is seqtk inside samtools? If not, where to get it? Thanks a lot!

[Istvan Albert]

You need to update samtools to at least 1.1 http://www.htslib.org/

Get seqtk from https://github.com/lh3/seqtk see also How To Obtain And Install Seqtk

[biolab]

Hi Istvan, your answer is really helpful.

Answer 3

[Chadi Saad]

You need to add '-' for seqtk in order to read from stdin :

samtools bam2fq input.bam | seqtk seq -A - > output.fa

[ATpoint]

You can add - but it is not necessary.

Answer 4

[GenoMax]

reformat.sh from BBMap suite will do the conversion efficiently.

reformat.sh in=your.bam out=seq.fa
reformat.sh in=your.sam out=seq.fa

Answer 5

[kristina.mahan]

samtools fasta input.bam > output.fasta

Answer 6

[Alex Reynolds]

Via bam2bed and bed2faidxsta.pl:

$ bam2bed < reads.bam | bed2faidxsta.pl > reads.fa

Or sam2bed:

$ sam2bed < reads.sam | bed2faidxsta.pl > reads.fa

• https://github.com/bedops/bedops (bam2bed or sam2bed)
• Gist for bed2faidxsta.pl

[xiaoleiusc]

I followed your instruction and even copied the unzipped bed2faidxstal.pl to the target folder, I got error "(base) bieniaszs-ipro:K20 bieniaszlab$ sam2bed < NL43_5562_filterredreads_header.sam | bed2faidxsta.pl > NL43_5562_filterredreads_header.fa -bash: bed2faidxsta.pl: command not found".

Then I add "perl" before the bed2faidxstal.pl, I still got error as below:

(base) bieniaszs-ipro:K20 bieniaszlab$ sam2bed < NL43_5562_filterredreads_header.sam | perl bed2faidxsta.pl > NL43_5562_filterredreads_header.fa
[E::fai_build3_core] Failed to open the file /Users/bieniaszlab/miniconda3/data/K_SRF11_U/K_20/K20/NL43_WT_CLIP.fa.gz
[faidx] Could not load fai index of /Users/bieniaszlab/miniconda3/data/K_SRF11_U/K_20/K20/NL43_WT_CLIP.fa.gz
[E::fai_build3_core] Failed to open the file /Users/bieniaszlab/miniconda3/data/K_SRF11_U/K_20/K20/NL43_WT_CLIP.fa.gz
[faidx] Could not load fai index of /Users/bieniaszlab/miniconda3/data/K_SRF11_U/K_20/K20/NL43_WT_CLIP.fa.gz
[E::fai_build3_core] Failed to open the file /Users/bieniaszlab/miniconda3/data/K_SRF11_U/K_20/K20/NL43_WT_CLIP.fa.gz
[faidx] Could not load fai index of /Users/bieniaszlab/miniconda3/data/K_SRF11_U/K_20/K20/NL43_WT_CLIP.fa.gz
[E::fai_build3_core] Failed to open the file /Users/bieniaszlab/miniconda3/data/K_SRF11_U/K_20/K20/NL43_WT_CLIP.fa.gz

[Alex Reynolds]

Put the directory containing the Perl script into your environment PATH, or run it with Perl. Make sure your FASTA is indexed with samtools.

[xiaoleiusc]

Thanks. I need FASTA output from SAM input, you mean to make sure my SAM file (not FASTA file) is indexed right?

[Alex Reynolds]

No, I mean that your reference genome FASTA must be indexed, in order to get out FASTA based on the BED coords that come out of the SAM conversion step. (Sorry for brevity, I was on a cell phone. Feel free to follow up if this isn't clear.)