Different info between "test_results" and "gff3"

[minnie0121]

• spladder version:3.0.2
• Python version:3.9.7
• Operating System:Linux

Description

I found the chromosome numbers and exon positions were different between the result of "build mode" (gff3) and "test mode" (tsv).
Considering the Ensembl Gene ID in each file is the same, the information in the gff3 is probably true.
Please let me know the cause of this difference and the effect for the test results.

"test_results_extended_C3_exon_skip.tsv"
event_id        chrm    exon_pos        alt_usage       gene_id gene_name
exon_skip.129637        7       151235047-151235628:151236407-151236795:151237373-151237593     0:1:0   ENSG00000012817 KDM5D

"merge_graphs_exon_skip_C3.confirmed.gff3"
Y       exon_skip       gene    19741318        19743239        .       -       .       ID=exon_skip.129637;GeneName="ENSG00000012817";HasNovelJunction="N"
Y       exon_skip       mRNA    19741318        19743239        .       -       .       ID=exon_skip.129637_iso1;Parent=exon_skip.129637;GeneName="ENSG00000012817";HasNovelJunction="N"
Y       exon_skip       exon    19741318        19741488        .       -       .       Parent=exon_skip.129637_iso1
Y       exon_skip       exon    19743162        19743239        .       -       .       Parent=exon_skip.129637_iso1
Y       exon_skip       mRNA    19741318        19743239        .       -       .       ID=exon_skip.129637_iso2;Parent=exon_skip.129637;GeneName="ENSG00000012817";HasNovelJunction="N"
Y       exon_skip       exon    19741318        19741488        .       -       .       Parent=exon_skip.129637_iso2
Y       exon_skip       exon    19741735        19741857        .       -       .       Parent=exon_skip.129637_iso2
Y       exon_skip       exon    19743162        19743239        .       -       .       Parent=exon_skip.129637_iso2

What I Did

I performed SplAdder "build mode" according to "Use on large cohorts" page (https://spladder.readthedocs.io/en/latest/spladder_cohort.html), then ran "test mode" using 3 samples for each group.

spladder test \
 --conditionA "${control_sample}" \
 --conditionB "${compare_sample}" \
 --labelA "${control_group}" \
 --labelB "${compare_group}" \
 --outdir "${spladder_dir}" \
 -v

[kevinigfl]

I obtain the same discrepancy, did you fix the bug?

[jamie-lyu]

Also I am suffering from the same issue.

[jbendik14]

I have been having the exact same issue. Did anyone ever find a solution/reason for this?

[kevinigfl]

Hi, I did not find the reason, but a solution is to take the event_id identified in the test files and go back to the build files where the positions (chr-start-end) are correct.
Hope it can help

[michael-kotliar]

Got the same problem. Completely unreliable tool if such simple errors occur.

[akahles]

Sorry for the long silence. I somehow managed not to get notifications on this repo 🤦 . In any case. The bug seems to have been introduced with the latest change set, where the output of coordinates was added to the testing results. Thus, as previous posters commented, matching via the event ID to the gff3 worked, but the coordinates output in the testing results have been mismatched. I'll fix this shortly.

Thank you to all who have been patient and replied with constructive feedback.

Best,
Andre

[akahles]

This should be fixed with the latest release. Please re-open if you find otherwise.