Finding novel transcripts

Question

[analyst]

Dear all,

I have run standalone blastx using default options for my list of transcripts and got output like this:

MSTRG.285.1 KAG7595701.1    68.976  332 9   3   194 1189    26  263 5.59e-112   342
MSTRG.285.1 KAG7595701.1    90.698  43  3   1   30  158 1   42  7.53e-13    81.3
MSTRG.285.1 CAD5311629.1    68.675  332 10  3   194 1189    26  263 5.14e-111   339
MSTRG.285.1 CAD5311629.1    90.698  43  3   1   30  158 1   42  6.85e-13    81.3
MSTRG.285.1 XP_020868191.1  67.683  328 54  3   206 1189    18  293 3.48e-107   331
MSTRG.285.1 KAG7591185.1    68.339  319 49  3   230 1186    39  305 3.10e-104   324
MSTRG.285.1 CAH8251728.1    67.812  320 51  3   230 1189    39  306 3.22e-103   321
MSTRG.285.1 KAG7653704.1    68.125  320 50  3   230 1189    39  306 3.33e-103   321
MSTRG.285.1 EFH68782.1  67.812  320 51  3   230 1189    38  305 3.44e-103   321
MSTRG.285.1 OAP14613.1  86.387  191 0   1   617 1189    92  256 5.16e-70    234

Please guide which filters should I apply to get novel transcripts?

Thanks

[GenoMax]

Define "novel". If your queries are showing good "hits" (like some above) to something in the database then they are not "novel" by definition of the word.

[analyst]

Good hits with minimum e-value right?

Should I take care of query coverage also? If yes please guide how can I calculate query coverage from above output file.

This is the command that I used:

blastx -query file.fa -db nr -strand plus -out output_blastx.txt -evalue 1E-10 -outfmt 6

[analyst]

Is there any source from where I can get idea which threshold or criteria to set for percentage identity parameter to get novel transcripts for example transcripts less than 80 or 90 represents novelty?

[biofalconch]

I encourage you to have a look at OrthoFinder: https://github.com/davidemms/OrthoFinder

Not only will you now which transcripts are new, you will get a full classification for every transcript

[analyst]

Thank you biofalconch!

I am looking for novel lncRNA transcripts from Arabidopsis thaliana data.

[biofalconch]

Thats a little bit harder, since lncRNAs usually tend to evolve differently, maybe something like what's proposed in this paper would be of use?

https://www.nature.com/articles/s41598-022-18254-0

They go through various filters to make sure they don't code for proteins, really not much trying to find lncRNAs using other species....