search

mbsohn / cmmb

Sparse compositional mediation model for binary outcomes
2 stars 1 forks source link

Microbiome data can't contain zeros #1

Open levlitichev opened 1 year ago

levlitichev commented 1 year ago

Hello,

I have been using your package CCMM for mediation analysis. I noticed a potential issue that may also apply to CMMB.

I posted about it in the CRAN page for CCMM: https://github.com/cran/ccmm/issues/1. In brief, I believe the code will fail if the microbiome data contains any zeros (when you take the log, e.g. here).

Just wanted to bring this to your attention in case it will benefit any future users.

Thanks, Lev

mbsohn commented 1 year ago

Dear Lev,

Thanks for contacting me about the zero issue. Function CCMM assumes no zero in a taxonomic profile. In the paper, we replaced zeros with 0.5 before transforming counts into proportions. Alternatively, imputed values can be used. Because of no consensus on which is better, we left that part to users. We are currently working on a systematic way of treating zeros and will include an option to use our new approach in CCMM in future.

Best Regard, Michael

From: Lev Litichevskiy @.> Date: Thursday, January 26, 2023 at 4:29 PM To: mbsohn/cmmb @.> Cc: Subscribed @.***> Subject: [EXT] [mbsohn/cmmb] Microbiome data can't contain zeros (Issue #1)

Hello,

I have been using your package CCMM for mediation analysis. I noticed a potential issue that may also apply to CMMB.

I posted about it in the CRAN page for CCMM: cran/ccmm#1https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_cran_ccmm_issues_1&d=DwMCaQ&c=4sF48jRmVAe_CH-k9mXYXEGfSnM3bY53YSKuLUQRxhCwrmQ9ZDKmVaAI6sBPw4Jo&r=RJT4Y3Ie0sD4zZzbhZScBOn_7aBN2b4lKsx2DA6xz08&m=p1WqTA7dL6ertieWScqIeaDxMjsghbn-7oWpAL8KQpkx_fovxXqXpcn5EJpQnk00&s=rk7vKhqY02CV3U69W2PKDE19TrmeBEAcZBsDRBnXUCo&e=. In brief, I believe the code will fail if the microbiome data contains any zeros (when you take the log, e.g. herehttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_mbsohn_cmmb_blob_b92a3b32e4e39b8ac82220e14ce8ea63bc0bccee_R_cmmb.R-23L92&d=DwMCaQ&c=4sF48jRmVAe_CH-k9mXYXEGfSnM3bY53YSKuLUQRxhCwrmQ9ZDKmVaAI6sBPw4Jo&r=RJT4Y3Ie0sD4zZzbhZScBOn_7aBN2b4lKsx2DA6xz08&m=p1WqTA7dL6ertieWScqIeaDxMjsghbn-7oWpAL8KQpkx_fovxXqXpcn5EJpQnk00&s=ChwM_dxf6qRL5v6oNjRUk-vhLwHeyQknw1kZC8_AuF8&e=).

Just wanted to bring this to your attention in case it will benefit any future users.

Thanks, Lev

— Reply to this email directly, view it on GitHubhttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_mbsohn_cmmb_issues_1&d=DwMCaQ&c=4sF48jRmVAe_CH-k9mXYXEGfSnM3bY53YSKuLUQRxhCwrmQ9ZDKmVaAI6sBPw4Jo&r=RJT4Y3Ie0sD4zZzbhZScBOn_7aBN2b4lKsx2DA6xz08&m=p1WqTA7dL6ertieWScqIeaDxMjsghbn-7oWpAL8KQpkx_fovxXqXpcn5EJpQnk00&s=x8LsOMyqOlKhdgsRJtc_zwO8mmQ3XKoUkd8LutaSXVI&e=, or unsubscribehttps://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AMEPBHEK6IMVUFO4MTMWKPLWULUBVANCNFSM6AAAAAAUH7NSZI&d=DwMCaQ&c=4sF48jRmVAe_CH-k9mXYXEGfSnM3bY53YSKuLUQRxhCwrmQ9ZDKmVaAI6sBPw4Jo&r=RJT4Y3Ie0sD4zZzbhZScBOn_7aBN2b4lKsx2DA6xz08&m=p1WqTA7dL6ertieWScqIeaDxMjsghbn-7oWpAL8KQpkx_fovxXqXpcn5EJpQnk00&s=iABbIFEJ6dOi1zp2k6zBTuY8ZsHmTqz-CdMjqGWQ8Nc&e=. You are receiving this because you are subscribed to this thread.Message ID: @.***>

levlitichev commented 1 year ago

Got it. Thank you for the clarification.