Exporting output data

Help & Support
1

Hey there,

I have been setting up a pipeline for variant prediction and would like to export the values underlying the track visualization.

However, I ran into a problem when trying to acess the RNA-Seq data, as there are multiple assays stored in output.reference.rnaseq and I don’t understand how to access/export only the values of one RNA-Seq assay at a time (Eg. I just want the values for total RNA-Seq on the positive strand).

Would really appreciate some help on this!

2

Hi There!

Thanks for reaching out.

For any given biosample you may have up to four output tracks for RNAseq: ‘polyA plus RNA-seq’ and ‘total RNA-seq’, each for both the positive and negative strand. To extract values for a specific RNA-seq assay and strand, you can use the filtering methods available on the TrackData object. First, isolate the strand using .filter_to_positive_strand(). Next, apply .filter_tracks() with a boolean mask on the .metadata DataFrame to select the desired assay (e.g., ‘total RNA-seq’). Finally, access the .values attribute to get the raw numpy array for exporting.

# 1. Filter to positive strand
pos_tracks = output.reference.rna_seq.filter_to_positive_strand()

# 2. Filter to specific assay
mask = pos_tracks.metadata['Assay title'] == 'total RNA-seq'
specific_track = pos_tracks.filter_tracks(mask.values)

# 3. Export values
exported_values = specific_track.values

Kind regards,
Tumi

3

Thank you so much, that helped a lot.
However, I ran into some trouble with the Chip Histone samples.
Since I only want to export the H3K27ac data, I tried this approach since there is no strand information in Chip Histone:

# 1. Filter to specific assay:
mask = output.reference.chip_histone.metadata["histone_mark"] == "H3K27ac"
specific_track = output.reference.chip_histone.filter_tracks(mask.values)

#2. Export
exported_values = specific.track.values

However I when exporting I get the error that the arrays must be of the same lenght. Which makes me believe that I am either not exporting the correct data or the interval input is wrong.

4

Hi There!

Thanks for the follow-up.

I havent been able to reproduce the error. After predicting your variant with this:

output = dna_model.predict_variant(
  interval=interval.resize(dna_client.SEQUENCE_LENGTH_1MB),
    variant=variant,
    requested_outputs=[dna_client.OutputType.CHIP_HISTONE],
    ontology_terms=target_curie,
)

The shape of the reference results will be (size of spatial axis, number of ChiP-histone subtypes)

output.reference.chip_histone.values.shape
(8192, 6)

You then apply your H3K27ac mask (small error in your code: . instead of _ in line 3)

mask = output.reference.chip_histone.metadata["histone_mark"] == "H3K27ac"
specific_track = output.reference.chip_histone.filter_tracks(mask.values)
exported_values = specific_track.values

Exported values now only holds H3K27ac data

exported_values.shape
(8192, 1)

Can you send me a code snippet that reproduces the error? Then we can also check your concerns about the input interval

Kind regards,
Tumi