transcript with very small EffectiveLength

[steven-bioinfo]

Hello,

I use salmon to analyse my metatranscriptomic data.

In my results I found, some very high TPM but this TPM have already a very small EffectiveLength. Because of this very small EffectiveLength, the TPM are very high (and less reproducible between my replicates).

How to analyse this TPM ? It's possible to remove this bias with a salmon argument ?

command line :

salmon quant -i {input.salmon_index} -l A -1 {input.fastq_1} -2 {input.fastq_2}  --validateMappings -o {params.output_dir} --meta -p {threads} --seqBias --gcBias

Exemple of Top3 TPM on a sample :

Name	Length	EffectiveLength	TPM	NumReads
lcl|NC_001136.10_cds_NP_010562.1_1187	168	8.000	27437.393774	1808.000
lcl|NC_001145.3_cds_NP_013978.1_4507	180	13.000	80351.033367	8604.000
lcl|NC_001137.3_cds_NP_010898.1_1506	132	1.000	85468.695651	704.000

Thanks,

Steven