HTseq reports missing attribute name

Question

[Bjorn]

Hello,

I am running this htseq command

htseq-count -r pos -t gene -i gene -s yes -f bam \ 
/Volumes/cachannel/ZebraFinchBrain/CB-4a_genomemapping/sorted_alignmentcb4a.bam \
/Volumes/cachannel/ZebraFinchBrain/GCF_003957565.2/Taeniopygia_guttata.bTaeGut1_v1.p.110.chr.gff3 >  \
/Volumes/cachannel/ZebraFinchBrain/HTSEQ_withautomate/output_counts.txt

However I get this error:

Error processing GFF file (line 75 of file /Volumes/cachannel/ZebraFinchBrain/GCF_003957565.2/Taeniopygia_guttata.bTaeGut1_v1.p.110.chr.gff3):
  Feature gene:ENSTGUG00000013637 does not contain a 'gene' attribute
  [Exception type: ValueError, raised in features.py:387]

But in my gff3 file there is clearly a gene attribute:

1   ensembl gene    34955   72180   .   +   .   ID=gene:ENSTGUG00000013637;Name=DCBLD2;biotype=protein_coding;description=discoidin%2C CUB and LCCL domain containing 2 [Source:NCBI gene%3BAcc:100218192];gene_id=ENSTGUG00000013637;logic_name=ensembl;version=2
1   ensembl mRNA    34955   72180   .   +   .   ID=transcript:ENSTGUT00000014203;Parent=gene:ENSTGUG00000013637;Name=DCBLD2-201;biotype=protein_coding;tag=Ensembl_canonical;transcript_id=ENSTGUT00000014203;version=2

Does anyone know what is going on here?

Answer 1

[Istvan Albert]

it is important to distinguish between the type column and an attribute

the attributes are in last column in the form attribute_name=attribute_value

using those definitions your line does not contain a gene attribute

it only contains a gene_id attribute (for what is worth that is probably what you need in the end)

[Bjorn]

Thanks! I was able to get around this error by using feature counts. However, I ended up getting zero reads. This is odd considering that my Bam file had an alignment of about 90%. I think that my .gff3 file and bam file are somehow incompatible.

Here is some of the .bam file. AS you can see it is region CM012081.2

J00138:157:HVCVCBBXX:4:1127:4919:41036  73      CM012081.2      207     60      135M    =       207     0       GGCGTCGCAGACCAAGAGCATAT
GAGTGCATTGGGGTGTTTGTCAGGCGCAATGTTAGAGCCCCATCTCCTCCCTACAGATCCTGTTCTAGTTTCAGTCACAAGGCCAGCCCCAGAGTGCCTGTGCCAGCCCCAG        AAFFFJJJJJJJJJJ
JJJJJJJJJJJFJJJJJJJJJJAJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJJFJJJJJJFJFJJJJJJJJJJJJJJJJJJJ7FJJF<JJJFJJJJJJJAFFJJF<JFJ<FJJJJJJFJ        AS:i:0

here is the head of my gff3 file with the alias almost matching the bam file except for the .1 instead of .2

1   ensembl gene    34955   72180   .   +   .   ID=gene:ENSTGUG00000013637;Name=DCBLD2;biotype=protein_coding;description=discoidin%2C CUB and LCCL domain containing 2 [Source:NCBI gene%3BAcc:100218192];gene_id=ENSTGUG00000013637;logic_name=ensembl;version=2
1   ensembl mRNA    34955   72180   .   +   .   ID=transcript:ENSTGUT00000014203;Parent=gene:ENSTGUG00000013637;Name=DCBLD2-201;biotype=protein_coding;tag=Ensembl_canonical;transcript_id=ENSTGUT00000014203;version=2
1   ensembl exon    34955   35066   .   +   .   Parent=transcript:ENSTGUT00000014203;Name=ENSTGUE00000142648;constitutive=1;ensembl_end_phase=1;ensembl_phase=0;exon_id=ENSTGUE00000142648;rank=1;version=2

Here is my output. I do not believe that the gene counts are all zero. This has been a sustained issue for 2 weeks and I am not sure what to do.

Geneid  Chr Start   End Strand  Length  /Volumes/cachannel/ZebraFinchBrain/CB-4a_genomemapping/sorted_alignmentcb4a.bam
gene:ENSTGUG00000013637 1   34955   72180   +   37226   0
gene:ENSTGUG00000013635 1   45219   302771  -   257553  0
gene:ENSTGUG00000020928 1   74112   115790  -   41679   0
gene:ENSTGUG00000013634 1   115954  334268  +   218315  0
gene:ENSTGUG00000027178 1   159330  160474  -   1145    0
gene:ENSTGUG00000013633 1   303920  334270  +   30351   0
gene:ENSTGUG00000013632 1   331197  351031  +   19835   0
gene:ENSTGUG00000013631 1   349213  372968  -   23756   0

[Istvan Albert]

your BAM file is mapped against a genome with chromosome called CM012081.2 but your annotation file has the chromosome called 1

the chromosomal names have to match

otherwise, it cannot match a gene to the the alignments

[Bjorn]

Thank you. I have been trying to build a genome for this GCF file but I keep running into the same error.

.fna genome >NC_044211.2 Taeniopygia guttata isolate Blue55 chromosome 1, bTaeGut1.4.pri, whole genome shotgun sequence AAGCCAGCCATATCAGTATCCCAGCCCGCAGGGATCTGACTGCCACCCAGCAATGGCCAGCCACTATGTGTGCCCCTGTA TCACT

however, I keep getting errors on during the hisat2-build command. That there are no A nucleotides, but when I open my .fna there clearly are. I haven't seen many examples of this on the internet. Any ideas?

Exited GFM loop fchr[A]: 0 fchr[C]: 306050481 fchr[G]: 526435265 fchr[T]: 746725375 fchr[$]: 1052636476 Exiting GFM::buildToDisk() Returning from initFromVector Wrote 355100291 bytes to primary GFM file: prof.1.ht2 Wrote 263159124 bytes to secondary GFM file: prof.2.ht2 Re-opening _in1 and _in2 as input streams Returning from GFM constructor assert_eq: expected (57130, 0xdf2a) got (57344, 0xe000) ./hgfm.h:2127 Assertion failed: (0), function HGFM, file ./hgfm.h, line 2127. Abort trap: 6

[Istvan Albert]

sounds like your genome file is not quite correct, you need to make sure it is in FASTA format all the way through, perhaps get it from a reliable source.

I tried to build the index on a valid data and it worked just fine:

# Install bio
pip install bio

# Fetch the accession number as fasta
bio fetch NC_044211 -format fasta > NC_044211.fa

# The indexing completes with no error
hisat2-build NC_044211.fa NC_044211