IgnoroMeNot outputs a list of ignorome genes highly associated with other well-annotated genes. Ignorome genes are genes that have little to no validated experimental Gene Ontology annotations (see this publication or this). Strong associations between ignorome genes and well-annotated genes can help fill the gaps in the protein function space and improve the balance in knowledge between "annotation-rich" and "annotation-poor" genes.
IgnoreMeNot is available on PyPI. To install, enter this command in Terminal:
pip install ignoromenot
IgnoroMeNot can be installed with conda. To install, use this command:
conda install -c anphan0828 ignoromenot
IgnoroMeNot requires a tab-separated file of a list of genes/gene products with any single annotation metric (e.g. GO annotation count, information content, article count). Highest and lowest annotated genes are defined based on this metric. Below is an example of how the rank table should look like:
| Genes | Metric |
|---|---|
| NUDT4B | 55.34890734396416 |
| HTT | 99.78875455312165 |
Ignoromenot also supports the calculation of three per-protein metrics: GO annotation count ('ct'), Wyatt Clark information content ('ia'), article count ('pmid'), fractional count ('fc'), if provided a Gene Ontology Annotation File (GAF). If a GAF is provided, users need to specify which metric should be used for ranking (refer to usage).
Additionally, a STRING interaction network file of the organism to be examined is required and can be downloaded from STRING. Example: E. coli interaction network. Note: large STRING files (e.g., human-id:9606) may cause IDE to crash.
usage: ignoromenot.py [-h] --input INPUT --stringppi STRINGPPI [--metric METRIC] [--aspect ASPECT] [--threshold_top THRESHOLD_TOP | --percentile_top PERCENTILE_TOP]
[--threshold_bot THRESHOLD_BOT | --percentile_bot PERCENTILE_BOT] [--threshold_ppi THRESHOLD_PPI | --percentile_ppi PERCENTILE_PPI]
IgnoroMeNot find ignorome genes
options:
-h, --help show this help message and exit
--threshold_top THRESHOLD_TOP, -ttop THRESHOLD_TOP
Set an absolute upper threshold for most annotated genes based on the given
metric. Default to 100
--percentile_top PERCENTILE_TOP, -ptop PERCENTILE_TOP
Set a relative upper threshold for most annotated genes at k-th percentile
based on the given metric. Cannot be provided simultaneously with
--threshold_top.Example: -ptop 95 selects top 5% of genes with highest value
--threshold_bot THRESHOLD_BOT, -tbot THRESHOLD_BOT
Set an absolute lower threshold for least annotated genes based on the given
metric. Default to 5
--percentile_bot PERCENTILE_BOT, -pbot PERCENTILE_BOT
Set a relative lower threshold for least annotated genes at k-th percentile
based on the given metric. Cannot be provided simultaneously with
--threshold_bot.Example: -pbot 10 selects top 10% of genes with lowest value
--threshold_ppi THRESHOLD_PPI, -tppi THRESHOLD_PPI
Set an absolute upper threshold for STRING protein-protein interaction score.
Default to 500
--percentile_ppi PERCENTILE_PPI, -pppi PERCENTILE_PPI
Set a relative upper threshold for STRING protein-protein interaction score.
Cannot be provided simultaneously with --threshold_ppi.Example: -pppi 95
selects top 5% of associated pairs with highest score
Specifiers:
--metric METRIC, -m METRIC
A single annotation metric obtained from GAF based on which the proteins will
be ranked. If not specified,proteins will be ranked based on annotation count.
Accepted metrics: ct, ic, ia. Default to 'ct'
--aspect ASPECT, -a ASPECT
If GAF file is provided as input, specify which aspect of GO to rank the
proteins. If not specified, proteins will be ranked based on the total value
across 3 aspects. Accepted aspects: All, MFO, BPO, CCO. Default to 'All' (sum
accross 3 aspects).
--recal RECAL, -r RECAL
Set recal to 1 if you are running on a new GAF, the program reconstructs the
ancestors and recalculates all metrics.Set recal to 0 if you only change the
threshold values but run on the same GAF, the program uses old files to save
time.Default to 1.
Required arguments:
--input INPUT, -i INPUT
Input could be a path to a tab-separated file of a list of genes with any
single annotation metric (see above sample rank table) or Input could be a GO
Annotation File (GAF). In this case, filename must end with .gaf
--stringppi STRINGPPI, -ppi STRINGPPI
The path to a STRING interaction network of the organism being examined. The
filename must start with the organism ID (e.g., 9606 for human, 511145 for
E.coli)
Demo data of E.coli are included in the GitHub repository.
$ ignoromenot --input demodata/WyattClarkIC-perprotein.tsv --idtable demodata/511145.protein.aliases.v11.5.txt --stringppi demodata/511145.protein.links.full.v11.5.txt --percentile_top 90 --percentile_bot 1 --threshold_ppi 850
This command reads 3 input files, where the genes coming from E.coli (511145) are ranked based on their Wyatt Clark information content (WyattClarkIC-perprotein.tsv). --percentile_top 90 indicates that the genes at the top 10% with respect to Wyatt Clark infromation content are taken,
--percentile_bot 1 takes the bottom 1% annotated genes based on Wyatt Clark information content (those are the ignorome genes), and --threshold_ppi 850 chooses STRING coexpression score (511145.protein.links.full.v11.5.txt) of 850 and above. The protein alias file (511145.protein.aliases.v11.5.txt) makes sure that protein names from different databases have their IDs mapped properly to STRING interaction network.
Input: GO annotation file of E.coli, from GO Consortium:
$ ignoromenot -i demodata/goa_eco27Jul22Filtered.gaf -id demodata/511145.protein.aliases.v11.5.txt -ppi demodata/511145.protein.links.full.v11.5.txt -m ct -a MFO -ptop 90 -pbot 1 -pppi 98
IgnoroMeNot outputs a list of ignorome genes based on these parameters. If IgnoroMeNot is run with internet connection, it also provides users with all STRING interaction partners of these ignorome genes above --threshold_ppi via STRING API.