Htseq unable to process last read

Question

[Aspire]

Counting reads with htseq-count produces the following error

Error occured when processing input (record #977823 in file <_io.TextIOWrapper name='<stdin>' mode='r' encoding='utf-8'>):
header not available in closed files [Exception type: ValueError, raised in libcalignmentfile.pyx:1884]

record #977823 is the last record (see below)

The command used to run htseq was

samtools view -F 2304 -f 1 -bh Samp1_PE.bam | samtools sort -n -O SAM - | htseq-count -s reverse -o Samp1_w_feature_PE.bam -p bam - /Annotation/Genes/masked.gtf >! htseq_pe.txt

The resultant text file is empty. The resultant bam file (Samp1_w_feature_PE) has all reads except for the last one (sorted by name). When counting the number of lines

> samtools view -F 2304 -f 1 -bh Samp1_PE.bam | wc -l

is exactly 977823, so I understand that record #977823 is the last record.

I have not found this error message when googling. Any help?

Answer 1

[Aspire]

Turns out that the specific parameter which causes this problem is

-o Samp1_w_feature_PE.bam -p bam

Counting paired-end reads AND creating the output in bam format causes an error. None of them separately causes an error. I was able to output bam, when counting single reads, and more importantly, I was able to output to sam format when counting paired end reads. I have converted the sam to bam at a later stage, so this is effectively a successful workaround.