In-frame indels shouldn't necessarily be "genes_inactivated"

[dannagifford]

When running gdtools COMPARE -f TSV, in frame deletions (here, Δ3 bp) are being recorded under the "genes_inactivated" column. I think "genes_inactivated" should only definitely apply to indels that generate a frameshift, i.e. mod(bp, 3)!=0

Maybe this was a deliberate choice--if so, it might be useful to mention in the docs somewhere.

The gene where I noticed this is an essential gene (gyrA), so I don't think it can be inactivated.

> MG1655_evo_muts %>% select(type, size, gene_name, genes_inactivated, genes_overlapping, genes_promoter) %>% filter(gene_name == "gyrA")

A tibble: 40 × 6
type size gene_name genes_inactivated genes_overlapping genes_promoter

1 SNP NA gyrA NA gyrA NA
2 DEL 3 gyrA gyrA NA NA
3 DEL 3 gyrA gyrA NA NA
4 DEL 3 gyrA gyrA NA NA
5 DEL 3 gyrA gyrA NA NA
6 DEL 3 gyrA gyrA NA NA
7 SNP NA gyrA NA gyrA NA
8 SNP NA gyrA NA gyrA NA
9 DEL 3 gyrA gyrA NA NA
10 DEL 3 gyrA gyrA NA NA

[jeffreybarrick]

Agree... looking into this logic.

[dannagifford]

I've also found that Δ1bp can also be called as "genes_overlapping" and sometimes "genes_inactivated" so I'm not really sure what overall is happening. Here, compare ompF and rapA. I've checked and the position of both mutations falls within the coding region. Both genes are on the reverse strand.

MG1655_mutations %>% filter(!is.na(size)) %>% select(type, size, position, gene_name, genes_inactivated, genes_overlapping, genes_promoter) %>% unique()

[dannagifford]

Hi Jeff,

I see you made some changes to this in 0.39.0 that address these--thanks! I do wonder if there's a possibility to still mark anything as inactivating if the insertion/deletion is out-of-frame, e.g. mod(size, 3)!=0 But this could get messy if e.g. the indel is small and toward the end of the protein, which might not be inactivating. There's no perfect solution really... and I think low priority as anyone with specific intentions is probably going to work it out themselves...

--inactivating-overlap-fraction
Mutations within this fraction of the length of a gene
from its beginning are assigned to the
'genes_inactivating' versus 'the genes_overlapping'
list if they fulfill other criteria. (DEFAULT=0.8)
--inactivating-size INS, DEL, and SUB mutations in genes that are longer
than this length cutoff are always assigned to the
'genes_inactivating' list, even if they are in-frame or
noncoding genes. (DEFAULT=15)

[jeffreybarrick]

Thanks for the ping here. I did not yet entirely address the logic here. It's a deep item on my TODO list. I agree that small in-frame indels in protein-coding genes should be excluded from the inactivating category by default.

[jeffreybarrick]

@dannagifford

I think I've fully addressed this. Now, INS/DEL/SUB are not marked as inactivating if they are near the end of a gene (defined via --inactivating-overlap-fraction ) or in-frame and smallish (≤ = --inactivating-size-cutoff ). The smallish INS/DEL/SUB are further checked to see if they introduce stop codons. If they do (and are not at the end of the gene), then they are also marked as inactivating.

The updated help describing this logic reads...

  --inactivating-overlap-fraction <arg> Mutations within
  this fraction of the length of a gene from its beginning
  are assigned to the 'genes_inactivating' versus 'the
  genes_overlapping' list if they fulfill other criteria.
  (DEFAULT=0.8)

  --inactivating-size-cutoff <arg> INS, DEL, and SUB
  mutations in genes that are longer than this length cutoff
  are always assigned to the 'genes_inactivating' list, even
  if they are in-frame or in noncoding genes. (DEFAULT=15)

  --promoter-distance <arg> Mutations upstream and within
  this distance of the beginning of a gene have it added to
  their 'genes_promoter' list. (DEFAULT=150)

...

MUTATION EFFECTS CLASSIFICATION

Each mutation has a 'genes_overlapping' list assigned based
on the genes it overlaps.

If the mutation affects a position within
--inactivating-overlap-fraction of the length of an
overlapping gene from its start, the gene is moved to the
'genes_inactivated' list if it is also a MOB, INV, INT, SNP
causing a nonsense mutation, or an INS, DEL, or SUB that
results in a size change that is <= the
--inactivating-size-cutoff and results in a frameshift or
stop codon in a protein-coding gene or an INS, DEL, SUB with
a size change > the --inactivating-size-cutoff for any type
of gene, even if it is in-frame in a protein-coding gene.

If there are no 'genes_overlapping' or 'genes_inactivated',
a mutation has a 'genes_promoter' list assigned to all genes
that are <= the --promoter_cutoff bp upstream of a gene.
There can be multiple qualifying genes assigned to each
list, but each gene will only be in one of the lists.

If this makes sense and sounds good, I can make a new release for you to test.

[cgliss]

@jeffreybarrick Hello, I am getting a similar issue where a 6bp in frame deletion and a 9 bp in frame insertion are being marked as "genes_inactivated". I'm using the most up to date breseq (0.39.0) and I've checked gdtools COMPARE and it does have the updated help in the March 7th comment above. Is this the correct call? I've attached the 2 genes result from gdtools COMPARE

BreseqInactivate.csv