modkit v0.6.1 filter commands not found

[EntoErin]

Hello,

When using modkit v0.6.1 (the most recent build available) to run pileup, both '--filter-threshold' and '--filter-percentile' commands are returning as not found. Has the ability to manually lower the threshold on pileup been removed? I cannot seem to find any information to suggest this.

Please can you advise?

[ArtRand]

Hello @EntoErin, could you paste the exact command you're using?

[EntoErin]

Hello, thanks for your prompt response!

I am using:

modkit pileup
input.bam
output.cpg.pileup.tsv
--cpg
--combine-strands
--combine-mods
--reference GCF_000001405.26_GRCh38_genomic.fna
--modified-bases C:m

and then I have tried both:

--filter-threshold 0.5

as well as

--filter-percentile 0.75

and

-p 0.75

and all of these filtering commands are returning as not found.

[ArtRand]

Hello @EntoErin,

Could you try this using the test data that is in the Modkit repo:

wget https://raw.githubusercontent.com/nanoporetech/modkit/refs/heads/master/tests/resources/CGI_ladder_3.6kb_ref.fa
wget https://github.com/nanoporetech/modkit/raw/refs/heads/master/tests/resources/bc_anchored_10_reads.sorted.bam
wget https://github.com/nanoporetech/modkit/raw/refs/heads/master/tests/resources/bc_anchored_10_reads.sorted.bam.bai

ref=CGI_ladder_3.6kb_ref.fa

# TODO: fill this in!
modkit=path/to/modkit

${modkit} pileup \
  bc_anchored_10_reads.sorted.bam \
  test_threshold.bed \
  --cpg \
  --combine-strands \
  --combine-mods \
  --reference ${ref} \
  --filter-threshold 0.5 \
  --modified-bases C:m

cat test_threshold.bed

${modkit} pileup \
  bc_anchored_10_reads.sorted.bam \
  test_percentile.bed \
  --cpg \
  --combine-strands \
  --combine-mods \
  --reference ${ref} \
  --filter-percentile 0.75 \
  --modified-bases C:m

cat test_percentile.bed

[EntoErin]

Hi @ArtRand

I have done this and it works which is great news, thanks for providing this.

I am no longer getting the 'command not found error' when I run the exact command you have shared with me using my reference .fa and my input bam file, however I am getting the following (specifically with the --filter-percentile at 0.75):

discarded 0 contigs with zero aligned reads
parsed 1 base modification(s), combining modification counts together for primary sequence bases 'C'
attempting to sample 10042 reads
Using filter threshold 0.9921875 for C
using optimised genome workers, combining strands at CpG motifs
~564723 failed processing
done, processed 0 rows

and therefore the output BED is empty. I've never encountered this issue before in previous iterations of modkit. The threshold being 0.9921875 seems rather large. If I set --filter-percentile to 0.5 it returns filter threshold 0.9667969.

The input bam file is ONT data which contains mixed modifications (C+m and C+h) in the same MM tag. Could this potentially be what is causing the issue?

Thank you again for your prompt responses, you've been really helpful!

[ArtRand]

Hello @EntoErin,

Could you try two things for me:

1. check-tags

modkit modbam check-tags $bam --head 10000

and paste the output
2. pileup without --filter-percentile or --filter-threshold?

[EntoErin]

Hi @ArtRand,

1. The output of check-tags:

checking tags on first 100000 reads
no errors
num PASS records: 100000 (100.00%)
num records: 100000
valid record tag headers:
+------------+-------+
| tag_header | count |
+------------+-------+
| C+h? | 99902 |
| C+m? | 99902 |
+------------+-------+

modified bases:
+--------+--------------+----------+------+
| strand | primary_base | mod_code | mode |
+--------+--------------+----------+------+
| + | C | h | ? |
| + | C | m | ? |
+--------+--------------+----------+------+

2. The output of pileup without --filter-percentile or --filter-threshold is as follows:
   discarded 0 contigs with zero aligned reads

parsed 1 base modification(s). Base modifications other than 'C:m' will be counted as 'N_other'.
attempting to sample 10042 reads
Threshold of 0.6308594 for base C is low. Consider increasing the filter-percentile or specifying a higher threshold.
using optimized genome workers, combining strands at CpG motifs
~564723 failed processing
Done, processed 0 rows.

[ArtRand]

Hello @EntoErin,

Ok, two more things :)

1. Run pileup with --log $log_file and send me the log file.
2. Email or attach a subset of the BAM (maybe one contig or region) that also fails to produce a pileup. You can email me at art.rand[at]nanoporetech.com

We'll get to the bottom of it. Thanks for the report.

[ArtRand]

Hello @EntoErin,

Thank you for sending the test data over. Pileup was failing due to a bug in v0.6.1 when you have modBAM records with Mm tags (instead of the more modern MM tags). This is a silly error on my part. I've attached a build with the fix. You can also use the following command on your modBAM files to make them compatible with v0.6.1 (the released version).

$ modkit modbam update-tags $bam updated.bam
$ samtools index updated.bam
$ modkit pileup updated.bam <..etc>

Thank you for reporting this bug, I'll have the fix in the next version.

modkit_v0.6.2_89c1dbd.zip

[EntoErin]

Hi @ArtRand,

Thank you very much for all of your help. I can confirm that this all works as it should now! Really appreciate you sorting this so fast.