[maubp]

Ah - blank lines at the end of the file can be easily overlooked.

In principle I believe that faidx can cope with blank lines *between* records, but I haven't made time to work out the required code changes to do this. It would probably save some general aggravation though

[NRP]

I had the same sort of segmentation fault. There were no blank lines in my file though. It was apparently due to one of my sequences being >65535 bp (though I don't know why that should matter).

Maubp's biopython code took care of it nicely. Thanks!

[skbrimer]

Just for anyone interested here is the Biopython equivalent:

from Bio import SeqIO
SeqIO.convert("inputfilename.fas", "fasta", "outputfilename.fas", "fasta")

The convert function returns the number of records if you wanted that information.

Thank you for this. Just ran into this problem.

[jmarshall]

$ ./samtools faidx /media/Poplar/baohua/genome/poplar_genome.fa
[fai_build_core] different line length in sequence 'scaffold_28'.

The problem with scaffold_28 in this file was the same problem that was reported here. It has now been fixed, and the fix will appear in samtools (and htslib) 1.3.

[jshaik]

Easiest way to fix this is to use seqret in emboss. Just convert the fasta file to ncbi style fasta and it automatically fixes the issue.